CN112308597B - Method for selecting facility address according to influence of sports user in competition environment - Google Patents
Method for selecting facility address according to influence of sports user in competition environment Download PDFInfo
- Publication number
- CN112308597B CN112308597B CN202010819078.1A CN202010819078A CN112308597B CN 112308597 B CN112308597 B CN 112308597B CN 202010819078 A CN202010819078 A CN 202010819078A CN 112308597 B CN112308597 B CN 112308597B
- Authority
- CN
- China
- Prior art keywords
- user
- influence
- candidate
- competition
- facility
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0204—Market segmentation
- G06Q30/0205—Location or geographical consideration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/248—Presentation of query results
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Databases & Information Systems (AREA)
- Accounting & Taxation (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Navigation (AREA)
Abstract
The invention discloses a method for selecting facility addresses according to influence on sports users in a competition environment, which comprises the steps of firstly, giving a plurality of candidate positions, and calculating a sports user set which can be influenced by each candidate position; adding the existing competition facilities to form competition with each candidate position, determining a sports user set which can be influenced by competition facilities and the competition of the candidate positions, and calculating competition influence values of a plurality of candidate positions according to the obtained sports user set; screening and sorting the candidate positions by using the obtained competition influence value to obtain a preferred position; according to the method, the strategy for optimizing the computational complexity is designed, so that the competitive site selection efficiency under a huge number of user scenes can be greatly improved, and the practicability of the method is further improved.
Description
Technical Field
The invention belongs to the technical field of space big data and data mining, and relates to a method for selecting a facility address according to influence on a sport user in a competition environment.
Background
The site selection with the greatest influence is an important space data mining problem, and has very wide application in the fields of business site selection, city planning and the like. Specifically, maximum impact location refers to mining an optimal location from a number of candidate addresses in a geographic space to build or deploy a facility so that the facility at that address can impact the most potential users.
With the increasing popularity of terminal devices with geographic positioning capability, conventionally, a user is regarded as a single stationary spatial location point, and a user group is regarded as an area density, which cannot objectively reflect the motion behavior characteristics of the user, for example, the user may shop near a residence or consume near a work place. Therefore, by comprehensively considering the influence of facilities on the whole user movement point set, the current maximum influence addressing technology (an addressing method and an apparatus ZL 201610347674.8) can incorporate the movement characteristics of the user into the addressing decision, and obtain more accurate addressing results than the traditional static addressing technology.
However, in real-world most influential addressing applications for moving users, ideal addressing environments that do not compete in the same row are often not present. For applications such as commercial store site selection, it is almost inevitable to face peer competition, i.e. facilities to be newly built are often required to compete with surrounding existing like facilities for potential sports users. Therefore, in a competition-oriented environment, the influence of the existing similar facilities on the users can obviously affect the influence evaluation of the newly built facilities, and neglecting competition relation factors among the facilities can reduce the accuracy and usability of the site selection result.
Disclosure of Invention
The invention aims to provide a method for selecting a facility address according to influence on a moving user in a competition environment, which solves the problem that the prior site selection technology aiming at the maximum influence of the moving user is difficult to apply to a peer competition scene because similar facility competition factors are not considered.
The technical scheme adopted by the invention is that the method for selecting the facility address according to the influence of the sports user in the competition environment is implemented according to the following steps:
step 1, a plurality of candidate positions are given, and a set of motion users which can be influenced by each candidate position is calculated;
step 2, adding the existing competition facilities to form competition with each candidate position, determining a sports user set which can be influenced by the existing facilities and the candidate positions under the competition condition, and calculating competition influence values of a plurality of candidate positions according to the obtained sports user set;
step 3, screening a plurality of candidate positions according to the competition influence value obtained in the step 2;
and step 4, sorting the candidate positions screened in the step 3 to obtain preferred positions.
The invention is also characterized in that:
the specific content of the step 1 comprises the following steps:
step 1.1, giving a plurality of candidate positions to obtain a candidate position set C, and obtaining any candidate position C in the candidate position set C i As Hash key, a triplet is establishedWherein element->Representation c i A set of mobile users that can be affected, initially an empty set; element->Representing for candidate position c i Initializing a user set to be determined as a user corpus;
step 1.2, traversing each motion user, calculating any motion user by using pruning rules based on a plurality of spatial motion points to obtain a subset of two candidate position sets which can necessarily influence and cannot influence the user, finding a corresponding triplet through a Hash value for each candidate position which can necessarily influence the user, and taking the user from elementsDeletion in the set while adding the user to +.>In the collection; for each candidate location that necessarily does not affect the user, the user is triad from the corresponding candidate location>Deleting in the collection;
step 1.3, after completing the traversal of all users, traversing all candidate locations and triples thereof, if candidate location c i In triplesThe set is empty, then +.>Namely, a user set which can be influenced by the candidate position; otherwise, it is required to follow the facility position c i For->The cumulative influence probability definition of each sports user is calculated, and the user capable of being influenced is added to +. >In the set, after traversing is completed, each candidate position obtains a user set which can be influenced by each candidate position;
the specific content of the step 2 comprises the following steps:
step 2.1 to move each user O in the user set O s Building user competition tuples as Hash key valuesOf (1), wherein->Is capable of affecting user o in all competing facilities s Is to be added to the competition facilities set of (a) for all users>Initializing an item as an empty set; build up to compete for facility f j Competition facility doublet as Hash keyWherein->The representation is directed to competing facilities f j To be determined, initially a complete set of users, when determining the competing facilities f j Can influence user o s When the user o s From the ∈two-tuple>Delete from the collection while f j Added to user o s Binary->In the set, a user set with the multi-energy influence of competing facilities is obtained;
step 2.2, according to candidate position c i User set of influenceThe accumulated computation c for each specific situation of the competing facilities affected by the user i Competing impact values of (2): by looking at +.>Each user o of (2) s (i.e.)>) User competitive binary group->Can determine c i At the right->For o in case of competition s Influence of (i.e.)>And then for all->The user in (c) can be obtained by accumulation i Competitive influence value->I.e. c i The sum of probabilities that a sports user can be affected by competing with existing facilities;
wherein in step 2.1 after obtaining the set of users each affected by the candidate location and the competing infrastructure, if there is one moving user o s Can be either the candidate position c i Influence, can be also competing by m (m is more than or equal to 1) facilities f 1 ，f 2 ，...，f m Case of influence, user accesses facility f 1 ，f 2 ，...，f m And c i The probability of f is the same 1 ，f 2 ，...，f m And c i Any facility to user o s The influence of (1/(m+1)) is 1;
wherein the accumulating part in the calculation of the competition effect value is considered in the step 2.1Of which only the calculation of c is required i User set of influences->Where "·" represents the calculation of the user influence or probability of influence when ignoring or considering the facility feature vector, independent of the candidate position c i Affected users can be pruned without traversing, so that the competitive dyadic group of the sports user affected by the competitive facility is calculated>When onlyIt is necessary to traverse a set of sports users affected by at least one candidate position, i.e. +.>Wherein C is a candidate location set;
wherein, the various features of different facilities in step 2.2 are not the same, and the competition influence value of the candidate position is calculated by using the facility weighting feature value according to the set weight of the different facility attribute requirements:
Let arbitrary facility f j (including candidate position c i Facility constructed at) forms a facility vector with features other than locationWherein->Representing the facility f to be considered in the application scenario of the present invention j After the feature scores are obtained, carrying out normalization calculation on a certain feature score of all facilities to obtain a feature vector of the end user: the normalized feature value of a specific facility is determined by dividing the feature score of the facility by the maximum feature score of all facilities, and the feature score can be defined by scoring according to specific application scenes, and the weight vector of the importance degree of the feature is w= (W) 1 ，w 2 ，...，w n ) Facility f j Is +.>At this time, when c i Influence user o s In the time, according to the competitive binary group->Can calculate c when considering the feature differences i To o s Influence of (i.e.)>Wherein candidate position c i Is +.>Thus, the candidate position c is considered in consideration of the feature differences of different facilities i The competition effect value of +.>
The specific content of the step 3 is as follows:
according to the competition influence values of all candidate positions obtained in the step 2, the optimal k candidate positions can be screened out;
the specific content of the step 3 is to introduce a candidate position competition influence value upper bound and a competition influence value threshold, and to screen by a pruning processing method, and the specific steps are as follows:
Step 3.1, according to the set of motion users affected by the candidate position obtained in step 1, making candidate position c i User set capable of being influencedThe size is +.>Since no competing influence of any competing facilities on the moving user is considered, for any user o affected by the candidate location s All have->I.e. < ->Candidate position c at this time i The influence exerted is maximum, and position c is selected when other facilities compete i Is not possible to exceed +.>Therefore, the influence of competition is not considered +.>For candidate position c i Upper bound of the competitive influence value of->By taking the upper bound of the competition influence value of the candidate positions as a key word, storing all the candidate positions in the Fibonacci large top heap HU, and rapidly determining k candidate positions with the maximum upper bound of the competition influence value;
calculating the competition influence value of the k 'candidate positions (k'. Gtoreq.k), wherein the k-th candidate position with the largest competition influence value is c k Then score (c) k ) Set as the competition effect value threshold score T For the remaining not yet calculated |C| -k' candidate positions, the upper bound of the competition effect value is smaller than the threshold score T Pruning is carried out on candidate positions of the competition influence values, the candidate positions of the competition influence values are calculated, the Fibonacci small top stacks hs taking the competition influence values as keywords are used for storage, and the initial small top stacks hs are empty stacks;
Step 3.2, stacking top candidate position c in the HU large top stack in step 3.1 (1) Ut Pop-up, wherein superscript (1) indicates the 1 st pop-up candidate position, then according to c (1) Ut User set of influenceFind and c (1) Ut Competitive facility, which first initializes the user competitive doublet +.>The set is empty, i.e. the two tuples without any user included therein, only the +.>Is calculated to influence the competing facilities of the user>For->Corresponding to the user inAnd +.>Feature vector of the competing facilities by calculating +.>Obtain->Competition influence value score (c) (1) Ut ) Then c (1) Ut Inserted as the first element into the small roof heap hs, a threshold score is set for the first time T The top element c is a small top stack hs st In (2) the competition effect value due to the fact that only c is present in hs (1) Ut An element, score T ＝score(c st )＝score(c (1) Ut )。
Step 3.3, treatment c according to step 3.2 (1) Ut Continues to pop up the top of stack candidate position c in the HU large top stack (2) Ut ，c (3) Ut ，...，c (i) Ut ，...，c (|C|) Ut I.e. 2 nd, 3 rd, … … th, i th, up to the last 1 pop-up candidate position;
step 3.4, processing the popped candidate positions of the HU large top stack one by one and updating the small top stack hs; at stack top candidate position c of pop-up large top stack HU (i) Ut When c (i) Ut Upper bound of competing influence value of (2)Less than a threshold score T Then know from the big top heap definition and the upper bound of the competing impact value +.>Will be ready to usec (i) Ut Pruning is carried out, and the rest candidate position c in the HU large roof pile (i+j) Ut Where j is an integer greater than 1 and i+j is less than or equal to |C|, pruning is performed on candidate positions after the ith element in the same way;
wherein in step 3.3 ifAnd user o s (i.e.)>) Has appeared in the user competitive binary group +.>In the set, the user is not calculated according to the user pruning method;
calculation ofObtaining a competitive influence value score (c) (i) Ut ) Inserting the small top heap hs in the step 3.2, if the number of candidates stored in the small top heap hs after the insertion is greater than k, ejecting the heap top element of the hs, namely only reserving k candidate positions with the largest competition influence value in the hs, and updating the heap top candidate position c 'in the small top heap hs after the updating' st Is updated to a threshold score T I.e. score T ＝score(c′ st ) Wherein c' st Representing new top of heap elements after each new insert element in hs and ejecting more than k elements, wherein the top of heap candidate position may not change when the number of elements in hs before insertion is less than k, then c' st ＝c st In other cases c' st Changes are necessarily made;
the specific content of the step 4 is as follows:
processing the candidate positions popped up by the HU large top heap one by one according to the step 3, and when the competition influence value upper bound of the candidate positions at the top of the heap is smaller than the threshold score T Or until the last element of the large top heap HU is popped up and processing is complete, k candidate positions in the small top heap hsThe k optimal candidate positions are obtained;
if the cost effectiveness of k candidate locations is considered, then a further selection can be made using the competing impact value of unit cost, for any of the optimal candidate locations c i If the cost of building a service facility at this location is cost (c i ) Then c can be calculated i Cost-effectiveness ratio, i.e. competitiveness, of (c)And calculating the competitiveness of the k optimal candidate positions based on the calculated competition force, and sequencing the k optimal candidate positions, wherein the maximum competitiveness is the optimal candidate position.
The invention has the beneficial effects that
Based on the space influence relation between the overall motility of the sports users and the facility positions, the invention analyzes the competition strength of different facilities on the users by considering the relative competition factors of the facilities, and realizes the mathematical modeling of the competition influence relation among the facilities, thereby quantifying the influence of the facility addresses on the sports objects in the competition environment, making an optimal site selection scheme facing the competition environment, avoiding the limitation that the existing site selection technology facing the sports users cannot be suitable for the competition environment, ensuring that the site selection decision result is more accurate and objective, and effectively improving the business and social benefits of the site selection of the facilities. By designing a strategy for optimizing the computational complexity, the competitive site selection efficiency under a huge number of user scenes can be greatly improved, and the practicability of the method is further improved.
Drawings
FIG. 1 is a schematic diagram of a competitive-site-oriented approach provided by an embodiment of a method of effecting a user selection of a facility address in a competitive environment in accordance with the present invention;
FIG. 2 is a schematic diagram of a user pruning process provided in accordance with an embodiment of a method of effecting a user selection of a facility address in a competitive environment in accordance with the present invention;
FIG. 3 is a schematic diagram of a candidate location pruning process provided in accordance with an embodiment of a method of effecting a user selection of a facility address in a competitive environment in accordance with the present invention;
fig. 4 (a), 4 (b), 4 (c) and 4 (d) are graphs showing the effects of embodiments of the present invention on changing the number of candidate locations, the number of sports users, the number of competing facilities and the k-value four-way aspect of the optimal candidates, respectively, according to the availability data of the prior art in a method of influencing the sports user's selection of a facility address in a competing environment.
Detailed Description
The invention will be described in detail below with reference to the drawings and the detailed description.
The invention provides a method for selecting facility addresses according to influence of sports users in a competitive environment, which is shown in fig. 1 and is implemented by the following steps:
step 1, firstly, according to a plurality of motion user positions in an application scene, determining motion users which can be influenced by each candidate position:
First, it is necessary to obtain movement point data of a moving user, which may be performed by, but not limited to, collecting movement trace data, point of interest geo-check-in data, life geo-log data, public transportation trip data, mobile communication location data, etc. of the user. It is noted that the present invention does not rely on a specific manner to obtain athletic data, and any data that may describe the user's geographic athletic performance location points may be used as athletic user athletic point data.
Step 1.1, according to the definition of the accumulated influence probability of a specific facility position on a plurality of space movement points of a single movement user, whether the facility position affects the movement user can be determined, and based on the definition, for a specific candidate position, a user set which can be affected by the candidate position can be obtained by traversing all the movement users and calculating the corresponding accumulated influence probability, wherein the size of the set is the influence of the candidate position. Performing the above traversal on all candidate positions to obtain the influence of the non-competition environment of each candidate position;
step 1.2, because the number of moving users is generally huge, the cost of computing the accumulated influence probability by traversing all the space moving points of each user is high, and the following method can be adopted for fast computing: first, using any candidate position C in the candidate position set C i As a Hash keyEstablishing triplesWherein element->Representation c i A user set capable of being influenced is initially an empty set; element->Representing for candidate position c i Is initialized to a user corpus. Each sports user is then traversed. For any moving user, a calculation using pruning rules based on multiple spatial movement points can result in a subset of the two candidate location sets that must be able to affect and not be able to affect the user. For each candidate position which necessarily can influence the user, the corresponding triplet is found by means of a Hash value, the user is taken from the element +.>Deletion in the set while adding the user to +.>In the collection; for each candidate location that necessarily does not affect the user, the user is triad from the corresponding candidate location>And deleting in the collection.
Step 1.3, after the traversal of all users is completed, the traversal is performed on all candidate positions and triples thereof. If candidate position c i In triplesThe set is empty, then +.>Namely, a user set which can be influenced by the candidate position; otherwise, needAccording to the facility position c i For->The cumulative influence probability definition of each sports user is calculated, and the user capable of being influenced is added to +. >In the collection. After the traversing is completed, each candidate position obtains a user set which can be influenced by each candidate position;
step 2, adding the same-row facilities to form competition with each candidate, determining competition facilities and candidate positions to compete for the sports users, and pruning calculation of irrelevant sports users in the process:
step 2.1 to move each user O in the user set O s Building user competition tuples as Hash key valuesOf (1), wherein->Is capable of affecting user o in all competing facilities s Is to be added to the competition facilities set of (a) for all users>Items are initialized to an empty set. In the case of not considering the influence of other competing facilities on the user, a similar method to the calculation of the candidate location influence of the step 0 can be used to calculate the user set that can be influenced by each competing facility, but there are two differences. The first difference is to establish to compete for the facility f j Competition facility doublet as Hash keyWherein->The representation is directed to competing facilities f j To be determined of a set of usersThe aggregate is initially a user corpus. Another difference from the candidate location influence calculation process is that when determining the competing facilities f j Can influence user o s When the user o s From the ∈two-tuple >Delete from the collection while f j Added to user o s Binary->In the collection;
after obtaining the set of users affected by the candidate location and the competing facilities, if there is one moving user o s Can be either the candidate position c i Influence, can be also competing by m (m is more than or equal to 1) facilities f 1 ，f 2 ，...，f m In the case of influence, the role of competition in influence calculation needs to be considered. Assume that competing facility f, except for geographic location, is not the same 1 ，f 2 ，...，f m And to be at candidate position c i The other features of the built or opened facilities are the same, namely the service content, the service quality, the scale, the price, the environment and the like of the facilities are all the same, and the user does not have any preference difference when selecting the service, so that the influence of the facilities on the user is equal at the moment, namely the user accesses the facility f 1 ，f 2 ，...，f m And c i The probability of f is the same 1 ，f 2 ，...，f m And c i Any facility to user o s The influence of (1/(m+1)) is 1;
due to the huge number of the moving users, the moving user set influenced by the computing competition facilities needs to traverse all the users, and the computing complexity is too high. Optionally, the pruning method based on candidate position influencing users can avoid traversing part of users, specifically considering accumulated part in competition influence value calculation Of which only calculation is neededc i User set of influences->Where "·" represents the calculation of the user influence or probability of influence when ignoring or considering the facility feature vector, then it is not subject to the candidate location c i Affected users can prune these users without traversing. Thus in calculating the competitive dyads of sports users affected by competitive facilities +.>For, only the set of sports users affected by at least one candidate position, i.e. +.>Wherein C is a candidate location set;
step 2.2, according to candidate position c i User set of influenceThe accumulated computation c for each specific situation of the competing facilities affected by the user i Is referred to as c in the present invention i Is a competitive influence value of (1). In particular, by looking at->Each user o of (2) s (i.e.)>) User competitive binary group->Can determine c i At the right->For o in case of competition s Influence of (i.e.)>And then for all->The user in (c) can be obtained by accumulation i Competitive influence value->I.e. c i The sum of probabilities that the sports users can be influenced by competing with facilities is basically different from the number of influencing users by influence statistics under the existing competition-free condition;
alternatively, the various features (or attributes) of different facilities are usually not identical, and the competition influence value of the candidate location can be calculated by using the facility weighting feature value according to the set weight of the different facility attributes; specifically, let arbitrary facility f j (including candidate position c i Facility constructed at) forms a facility vector with features other than locationWherein->Representing the facility f to be considered in the application scenario of the present invention j The values of the n features can be obtained from traditional approaches such as network evaluation information, user investigation and the like, or can be obtained through machine learning and other modes capable of obtaining facility feature scores, and after the feature scores are obtained, normalization calculation is carried out on certain feature scores of all facilities to obtain the feature vector of the invention of the end user. Specifically, the normalized feature value of a specific facility is determined by dividing the feature score of the facility by the maximum feature score of all facilities, and the feature score can be defined according to the specific application scenario. The demand importance weight vector for the feature is w= (W) 1 ，w 2 ，...，w n ) Facility f j Is given by the weighted eigenvalue of (2)At this time, the liquid crystal display device,when c i Influence user o s In the time, according to the competitive binary group->Can be calculated under consideration c when the features are different i To o s Influence of (i.e.)>Wherein candidate position c i Is +.>Thus, the candidate position c is considered in consideration of the feature differences of different facilities i The competition effect value of +.>
Step 3, verifying candidates which have the highest potential of becoming optimal positions one by one on the basis of the competition influence relation, and pruning poorer candidate addresses by using a threshold value of the competition influence value:
Given the cost effectiveness at candidate locations, i.e., the balance between the cost of building or provisioning a service at that location and the competing impact values, it is better to give k (k is an integer greater than 1) optimal candidate locations than just 1 optimal location. According to step 2, calculating competition influence values of all candidate positions, namely selecting optimal k candidate positions, wherein the process needs to process whether a large number of competition facilities influence the calculation of users, so that, optionally, the invention designs a candidate position pruning processing method capable of remarkably reducing the calculation complexity by utilizing a competition influence value threshold according to the upper bound of the competition influence values of the candidate positions, and the method is described in detail below:
step 3.1, according to the calculation in step 1, letting the candidate position c i User set capable of being influencedThe size is as followsSince no competing influence of any competing facilities on the moving user is considered, for any user o affected by the candidate location s All have->I.e. < ->Candidate position c at this time i The impact exhibited is greatest. Selecting position c when there is other facilities competing i Is not possible to exceed +.>Therefore, the influence of competition is not considered +.>For candidate position c i Upper bound of the competitive influence value of->By taking the upper bound of the competition influence value of the candidate positions as a key word, storing all the candidate positions in the Fibonacci large top heap HU, and rapidly determining k candidate positions with the maximum upper bound of the competition influence value;
assuming that the competition effect value of k 'candidate positions (k'. Gtoreq.k) is obtained by calculation, the candidate position with the k-th largest competition effect value is c k Then score (c) k ) Set as the competition effect value threshold score T . For the remaining uncomputed |C| -k' candidate positions, the upper bound of the competing impact values is less than the threshold score T I.e. the candidate positions whose competing influence in the case of competition is not likely to be higher than the top k candidate positions for which the calculated value of the competing influence is the largest, the pruning process can be performed directly. The candidate position of the competition influence value is calculated, a Fibonacci small top heap hs taking the competition influence value as a key word is used for storage, and the initial small top heap hs is an empty heap;
alternatively, other types of heap data structures may be used as alternative storage structures for the large top heap HU and the small top heap hs.
Step 3.2, top of Stack candidate position c in the HU Large Top Stack (1) Ut Pop-up, wherein superscript (1) indicates the 1 st pop-up candidate position, then according to c (1) Ut User set of influenceFind and c (1) Ut Competing facilities. This procedure first initializes the user competitive two-tuple +.>The set is empty, i.e. contains no tuples of any users. According to the sport user pruning method of step 2, only +.>Is calculated to influence the competing facilities of the user>For->Corresponding user->And +.>Feature vector of the competing facilities by calculating +.>Obtain->Competition influence value score (c) (1) Ut ) Then c (1) Ut Inserted as a first element into the small roof heap hs.Setting a threshold score for the first time T The top element c is a small top stack hs st In (2) the competition effect value due to the fact that only c is present in hs (1) Ut An element, score T ＝score(c st )＝score(c (1) Ut )；
Step 3.3, according to the above treatment c (1) Ut Similar procedure continues to pop up the top of stack candidate position c in the HU large top stack (2) Ut ，c (3) Ut ，...，c (i) Ut ，...，c (|C|) Ut (i.e., 2 nd, 3 rd, … … th, i th, up to the last 1 pop-up candidate), but the process is three different. The first difference is that ifUser o in (a) s (i.e.)>) Has appeared in the user competitive binary group +.>In the set, the user is not calculated any more according to the user pruning idea. The second difference is that +.>Obtaining a competitive influence value score (c) (i) Ut ) And inserting a small top heap hs, if the number of candidates stored in the small top heap hs after insertion is greater than k, ejecting a heap top element of the hs, namely only k candidate positions with the largest competition influence value are reserved in the hs. The third difference is that the top candidate position c 'in the updated small top stack hs needs to be updated' st Is updated to a threshold score T I.e. score T ＝score(c′ st ) Wherein c' st Representing new top of heap elements after each new insert element in hs and ejecting more than k elements, wherein the top of heap candidate position may not change when the number of elements in hs before insertion is less than k, then c' st ＝c st In other cases c' st Changes are necessary.
Step 3.4, processing the popped candidate positions of the HU large top stack one by one and updating the small top stack hs; at stack top candidate position c of pop-up large top stack HU (i) Ut It is not to pop up to the last element. When c (i) Ut Upper bound of competing influence value of (2)Less than a threshold score T Then know +.>I.e. c (i) Ut Must not become the optimal k candidate positions, and pruning can be performed. Further, for the remaining candidate position c in the HU large top heap (i+j) Ut Where j is an integer greater than 1 and i+j is less than or equal to |C|, then pruning can be performed on candidate positions subsequent to the ith element in a similar manner.
Step 4, through the processing of the process, the k optimal candidate positions are selected out as the most competitive results:
processing the candidate positions popped up by the HU large top heap one by one according to the step 3, and when the competition influence value upper bound of the candidate positions at the top of the heap is smaller than the threshold score T Or until the last element of the large top stack HU is popped up and the processing is completed, the k candidate positions in the small top stack hs are the optimal k candidate positions;
optionally, calculating competition influence values of all candidate positions according to the step 2, and selecting optimal k candidate positions;
alternatively, if cost effectiveness of k candidate locations needs to be considered, further selections may be made using the competing impact value of unit cost. Specifically, for any optimal candidate position c i If the cost of building a service facility at this location is cost (c i ) Then c can be calculated i Cost-effectiveness ratio, i.e. competitiveness, of (c)And calculating the competitiveness of the k optimal candidate positions based on the calculated competition force, and sequencing the k optimal candidate positions, wherein the maximum competitiveness is the optimal candidate position.
The designed competition influence probability calculation method can quantify the overall competition influence value of the candidate position competition environment on a plurality of sports users when the candidate position and the competition facilities influence the sports users simultaneously. The competition impact value may be further adapted to take into account the impact of other features of the facility on competition in addition to the location attribute by means of a facility feature vector normalization weighting calculation. In addition, considering the calculation complexity caused by the huge quantity of the moving users, according to the candidate position influence users, the influence calculation of partial users and competition facilities can be pruned, and meanwhile, the calculation of a large quantity of poor candidate position competition influence values can be pruned by utilizing the upper boundary of the candidate position competition influence values and the competition influence value threshold value. According to the above method, it is further possible to combine the cost benefits of k candidate locations to obtain an optimal addressing scheme in a competing environment.
In the invention, when a plurality of facilities (including facilities to be built on candidate positions and competing facilities) can influence a specific sport user at the same time, the user selects which facility to acquire service without considering the characteristic difference outside the facility positions, namely the influence probability of the facilities on the user is the same, and when the characteristic difference of each facility is considered, the influence probability of the facilities on the user is characterized by utilizing the magnitude of the facility weighting characteristic value according to the weight contributed by different attributes of the facilities when competing the user. The facility competition influence model is analogous to a webpage jump model in a webpage search classical algorithm PageRank (the probability of clicking is the same if the links on the webpage have no preference difference for users, and the probability of clicking is higher if the links on the webpage are relatively important for users), and the facility competition influence model is also applicable to practical application by analogous the links in the webpage, the links clicked and the link importance to facility, user selection facility and facility weighting characteristic values;
On the other hand, according to step 2, firstly, the number of competition facilities related to the candidate location influence user set is calculated, then, competition influence values of all candidate locations are calculated, and finally, the optimal k site selection results are obtained, wherein the calculation complexity is |C|O|F|, wherein |C| is the number of candidate locations, |O| is the number of moving users, and |F| is the number of competition facilities. The calculation complexity is reduced to |C ' ||O ' |F| by using two processing modes of the irrelevant users of the pruning part in the step 2 and the worse candidate positions of the pruning in the step 3, wherein O ' is a user set influenced by the candidate positions, namely O ' E O, and the irrelevant users of the pruning part lead |O ' | < |O|; c 'is the number of candidates actually needed to be calculated after pruning of the poor candidate positions is achieved by using the large top heap HU and the small top heap hs, and because the user activities are normally distributed in a biased manner, the number of candidates is |C' | < |C|. In conclusion, through two types of pruning treatment, the site selection based on the candidate position competition influence value can be more efficiently carried out;
examples
In view of the fact that the invention performs the site selection according to the competition influence value of the candidate position, the site selection scheme aiming at the influence probability of the sports user in the current temporary non-competition environment can only perform the efficiency comparison with the existing maximum influence site selection technology (an site selection method and device ZL 201610347674.8) of the sports user, and the comparison also needs to be adapted according to the competition influence value calculation: and (2) respectively calculating all candidate positions and the motion users which can be influenced by the competition facilities, traversing all candidate positions, calculating respective competition influence values according to the competition influence value calculation method provided by the invention, and taking k candidate positions with the largest competition influence values as the optimal site selection scheme. Both methods of comparison are implemented using the same language, run on the same platform and employ the same dataset; as shown in fig. 4 (a), 4 (b), 4 (c) and 4 (d), the method proposed by the present invention is significantly illustrated to be more efficient in terms of changing the number of candidate locations, the number of sports users, the number of competing facilities and the k value of the optimal candidate, respectively.
Taking the candidate location in FIG. 2 as an example, assuming that the candidate location relationship that must/cannot affect the user and that requires further verification is available using pruning is as in Table 1, the candidate location tripletThe fast calculation process of (2) is as follows: for o 1 And o 2 C can be obtained by using pruning rules 1 Necessarily affect, o 1 And o 2 Insert c 1 Triplet->And from->And delete, c 2 And c 3 Necessarily have no influence, then directly go o 1 And o 2 From c 2 ，c 3 Is>And->Delete in the middle; for o 3 Only c can be obtained by using pruning rules 3 Necessarily do not affect, o 3 From c 3 Is>Deleted in (c), and then is verified and calculated to obtain c 1 Can influence o 3 And c 2 Will not affect, o 3 Insert c 1 Triplet->And from->Delete o 3 From c 2 Is>And deleted. Similarly, o can continue to be calculated 4 To o 8 Continuously updating->And->Finally, c can be obtained 1 ，c 2 ，c 3 Of (2), wherein->
TABLE 1
As shown in fig. 2, where the arrow indicates the candidate location and the sports user that the competing infrastructure can affect: in candidate position c 2 For example, it can be seen thatTherefore, according to the pruning method of the candidate position influencing the user, only the user o needs to be traversed 4 ，o 5 Without considering the other 6 users. For user o 4 ，o 5 The competing facilities that can influence them are +. >And->Then c 2 Calculation of the Competition influence value is +.>It is to be noted that, due to the candidate position c 1 ，c 2 ，c 3 None can affect the sports user o 7 And o 8 (sports user contained in the shaded portion in FIG. 2), thus competing for the doublet +.>When it is unnecessary to calculate o 7 And o 8 For this example, the complexity drops to 3/4 (all users |o|=8 need not be calculated with pruning, and |o' |=6 after pruning);
if the effect of the facility feature vector is considered, the competition impact value calculation process is similar. In candidate position c 2 For example, the eigenvalues of the relevant facilities are shown in table 2, assuming that the facility eigenvectors have two dimensions: user score (0 star to 5 star) and service environment (preferably: 5 minutes, good: 4 minutes, medium: 2.5 minutes, bad: 1 minute), weight vector of these two dimensions is w= (W) 1 ，w 2 ) = (0.6,0.4), then c 2 Weighted eigenvalues of (2)Similarly, f 1 And f 2 The weighted eigenvalues of (2) are +.>And->Thus c 2 The competition effect value of +.>
TABLE 2
Using the influence relationship shown in FIG. 2, the processing method of pruning candidate positions using the threshold of the competing influence value is illustrated, with k being 2, as shown by the large top heap HU on the left side in FIG. 3, the candidate positionsOrdering from big to small according to the upper bound key of the respective competition value. First, the top candidate position c 1 Pop up from the large top heap HU, where two elements c remain in the large top heap HU 2 ，c 3 The element at the top of the stack is c 2 . Due toCan obtain their corresponding user competition binary groups, can obtain Further, a competition effect value score (c) 1 ) =1.5, then it is inserted into the small roof pile hs, where hs shown in the upper right corner of fig. 3 is only c 1 An element, thus setting a threshold score T ＝score(c 1 ) The method comprises the steps of carrying out a first treatment on the surface of the Then pop up the heap top element in HU as c 2 . Due to->Then it can be obtainedAnd->And then get c 2 Competition influence value score (c) 2 ) =1.33. At c 2 After inserting the small top heap hs, the small top heap stores exactly k=2 candidate positions. As shown in the lower right corner of fig. 3, due to score (c 1 )＞score(c 2 ) So the candidate position c 2 Put on top of the stack. At this time, the update threshold is the competition effect value score of the heap top element T ＝score(c 2 ). At this time, due to the stack top candidate position c in the large top stack HU 3 Upper bound of the competitive influence value of->Less than a threshold score T =1.33, then the large can be takenAll remaining candidate positions in the top stack HU are pruned, i.e. no calculation of c is required 3 Is a competitive influence value of (1). Finally, the candidate position subset { c 1 ，c 2 And k=2.
For the case where the cost effectiveness of the k best candidate locations is considered, the data of table 1, table 3, is taken as an example for illustration. Calculating respective cost-benefit ratios based on the competing impact values and costs of the candidate locations, i.e Based on this, candidate position c can be determined 2 Is the most cost effective and optimal result.
TABLE 2
| c 1 | c 2 | c 2 | |
| Competition impact value score (c) i ) | 1.5 | 1.33 | 0.5 |
| Construction cost (c) i ) | 10 ten thousand yuan | 8 ten thousand yuan | 6 ten thousand yuan |
Claims (11)
1. A method for selecting a facility address in a competitive environment based on influencing sports users, characterized by the steps of:
step 1, determining whether a facility position affects a motion user according to the definition of accumulated influence probability of a specific facility position on a plurality of spatial motion points of a single motion user, and based on the definition, traversing all the motion users and calculating corresponding accumulated influence probabilities for a specific candidate position to obtain a user set which can be affected by the candidate position, wherein the size of the user set is the influence of the candidate position, and traversing all the candidate positions to obtain the influence of each candidate position in a non-competitive environment;
step 2, adding the existing competition facilities to form competition with each candidate position, determining a sports user set which can be influenced by the existing facilities and the candidate positions under the competition condition, calculating pruning irrelevant sports users in the process, and calculating competition influence values of a plurality of candidate positions according to the obtained sports user set;
Step 3, according to the competition influence value obtained in the step 2, introducing a competition influence value upper bound and a competition influence value threshold of the candidate positions, and screening a plurality of candidate positions by a pruning processing method;
and step 4, sorting the candidate positions screened in the step 3 to obtain preferred positions.
2. A method for selecting a facility address based on influencing sports users in a competitive environment as defined in claim 1 wherein said step 1 details include:
step 1.1, giving a plurality of candidate positions to obtain a candidate position set C, and obtaining any candidate position C in the candidate position set C i As Hash key, a triplet is establishedWherein element->Representation c i A set of mobile users that can be affected, initially an empty set; element->Representing for candidate position c i Initializing a user set to be determined as a user corpus;
step 1.2, traversing each motion user, calculating any motion user by using pruning rules based on a plurality of spatial motion points to obtain a subset of two candidate position sets which can necessarily influence and cannot influence the user, finding a corresponding triplet through a Hash value for each candidate position which can necessarily influence the user, and taking the user from elements Deletion in the set while adding the user to +.>In the collection; for each candidate location that necessarily does not affect the user, the user is triad from the corresponding candidate location>Deleting in the collection;
step 1.3, after completing the traversal of all users, traversing all candidate locations and triples thereof, if candidate location c i In triplesThe set is empty, then +.>Namely, a user set which can be influenced by the candidate position;otherwise, it is required to follow the facility position c i For->The cumulative influence probability definition of each sports user is calculated, and the user capable of being influenced is added to +.>And in the set, after the traversal is completed, each candidate position obtains a user set which can be influenced by each candidate position.
3. A method for selecting a facility address based on influencing sports users in a competitive environment as defined in claim 1 wherein said step 2 details include:
step 2.1 to move each user O in the user set O s Building user competition tuples as Hash key valuesOf (1), wherein->Is capable of affecting user o in all competing facilities s Is to be added to the competition facilities set of (a) for all users>Initializing an item as an empty set; build up to compete for facility f j Competitive facility binary set as Hash key>Wherein->The representation is directed to competing facilities f j To be determined, initially a complete set of users, when determining the competing facilities f j Can influence user o s When the user o s From the ∈two-tuple>Delete from the collection while f j Added to user o s Of two groupsIn the set, a user set which can be influenced by competing facilities is obtained;
step 2.2, according to candidate position c i User set of influenceThe accumulated computation c for each specific situation of the competing facilities affected by the user i Competing impact values of (2): by looking at +.>Each user o of (2) s User competitive binary group->Determining c i At the right->For o in case of competition s Influence of (i.e.)>And then for all->The user in (c) can be obtained by accumulation i Competitive influence value->I.e. c i The sum of probabilities of moving users can be influenced by competing with existing facilities.
4. A method of selecting a facility address based on influencing a sports user in a competitive environment according to claim 3, after the acquisition of the set of users affected by the candidate location and the competing infrastructure in step 2.1, if there is a moving user o s Can be either the candidate position c i Influence, can be also competing by m (m is more than or equal to 1) facilities f 1 ，f 2 ，...，f m Case of influence, user accesses facility f 1 ，f 2 ，...，f m And c i The probability of f is the same 1 ，f 2 ，…，f m And c i Any facility to user o s The influence of (2) is 1/(m+1).
5. A method for selecting a facility address based on impact of a moving user in a competitive environment as claimed in claim 3 wherein said step 2.1 considers the accumulated portion of the competition impact value calculationOf which only the calculation of c is required i User set of influences->Where "·" represents the calculation of the user influence or probability of influence when ignoring or considering the facility feature vector, independent of the candidate position c i Affected users can be pruned without traversing, so that the competitive dyadic group of the sports user affected by the competitive facility is calculated>When it is only necessary to traverse the set of sports users affected by at least one candidate position, i.e. +.>Where C is the candidate set of locations.
6. A method for selecting a facility address according to the influence of a sports user in a competitive environment according to claim 3, wherein the various features of the different facilities in step 2.2 are not identical, and the facility weighted feature value is used to calculate the competition influence value of the candidate location according to the set weight for the different facility attribute requirements: let include candidate position c i Arbitrary facility f of facilities constructed at site j Features other than location constitute facility vectorsWherein->Representing facility f to be considered in application scenario j After the feature scores are obtained, carrying out normalization calculation on a certain feature score of all facilities to obtain a feature vector of the end user: the normalized feature value of a specific facility is determined by dividing the feature score of the facility by the maximum feature score of all facilities, and the feature score can be defined by scoring according to specific application scenes, and the weight vector of the importance degree of the feature is w= (W) 1 ，w 2 ，...，w n ) Facility f j Is given by the weighted eigenvalue of (2)At this time, when c i Influence user o s In the time, according to the competitive binary group->Can calculate c when considering the feature differences i To o s Influence of (i.e.)>Wherein candidate position c i Is +.>Thus, the candidate position c is considered in consideration of the feature differences of different facilities i The competition effect value of +.>
7. A method for selecting a facility address based on influencing sports users in a competitive environment as set forth in claim 3 wherein said step 3 is specifically defined as:
and (3) screening out optimal k candidate positions according to the competition influence values of all the candidate positions obtained in the step (2).
8. A method for selecting a facility address in a competitive environment based on influencing sports users according to claim 3, wherein said step 3 is specifically implemented as follows:
step 3.1, according to the set of motion users affected by the candidate position obtained in step 1, making candidate position c i User set capable of being influencedThe size is +.>Since no competing influence of any competing facilities on the moving user is considered, for any user o affected by the candidate location s All have->I.e. < ->Candidate position c at this time i The influence exerted is maximum, and position c is selected when other facilities compete i Is not possible to exceed +.>Therefore, the influence of competition is not considered +.>For candidate position c i Upper bound of the competitive influence value of->By taking the upper bound of the competition influence value of the candidate positions as a key word, storing all the candidate positions in the Fibonacci large top heap HU, and rapidly determining k candidate positions with the maximum upper bound of the competition influence value;
when calculating the competition influence value of k 'candidate positions (k'. Gtoreq.k), the candidate position with the largest competition influence value is c k Then score (c) k ) Set as the competition effect value threshold score T For the remaining not yet calculated |C| -k' candidate positions, the upper bound of the competition effect value is smaller than the threshold score T Pruning is carried out on candidate positions of the competition influence values, the candidate positions of the competition influence values are calculated, the Fibonacci small top stacks hs taking the competition influence values as keywords are used for storage, and the initial small top stacks hs are empty stacks;
step 3.2, stacking top candidate position c in the HU large top stack in step 3.1 (1) Ut Pop-up, where superscript (1) indicates the 1 st pop-up candidate location, then according to c (1) Ut User set of influenceFind and c (1) Ut Competitive facility, which first initializes the user competitive doublet +.>The set is empty, i.e. contains no tuples of any user, onlyIs calculated to influence the competing facilities of the user>For->Corresponding to the user inAnd +.>Feature vector of the competing facilities by calculating +.>Obtain->Competition influence value score (c) (1) Ut ) Then c (1) Ut Inserted as the first element into the small roof heap hs, a threshold score is set for the first time T The top element c is a small top stack hs st In (2) the competition effect value due to the fact that only c is present in hs (1) Ut An element, score T ＝score(c st )＝score(c (1) Ut )；
Step 3.3, treatment c according to step 3.2 (1) Ut Continues to pop up the top of stack candidate position c in the HU large top stack (2) Ut ，c (3) Ut ，...，c (i) Ut ，...，c (|C|) Ut I.e. 2 nd, 3 rd, … … th, i th, up to the last 1 pop-up candidate position;
Step 3.4, processing the popped candidate positions of the HU large top stack one by one and updating the small top stack hs; at stack top candidate position c of pop-up large top stack HU (i) Ut When c (i) Ut Competing shadow of (C)Upper bound of sound valueLess than a threshold score T Then know from the big top heap definition and the upper bound of the competing impact value +.>I.e. pair c (i) Ut Pruning is carried out, and the rest candidate position c in the HU large roof pile (i+j) Ut Where j is an integer greater than 1 and i+j is less than or equal to |C|, then pruning can be performed on candidate positions subsequent to the ith element in a similar manner.
9. A method for selecting a facility address based on influencing sports users in a competitive environment as set forth in claim 8 wherein if in step 3.3User o in (a) s Has appeared in the user competitive binary group +.>In the collection, < >>The user is not calculated according to the user pruning method;
calculation ofObtaining a competitive influence value score (c) (i) Ut ) Inserting the small top heap hs in the step 3.2, if the number of candidates stored in the small top heap hs after the insertion is greater than k, ejecting the heap top element of the hs, namely only reserving k candidate positions with the largest competition influence value in the hs, and updating the heap top candidate position c 'in the small top heap hs after the updating' st Is updated to a threshold score T I.e. score T ＝score(c′ st ) Wherein c' st Representing new top-of-heap elements after each new insert element in hs and ejecting more than k elements, wherein the top-of-heap candidate position is unchanged when the number of elements in hs before insertion is less than k, then c' st ＝c st In other cases c' st Changes are necessary.
10. A method for selecting a facility address based on influencing sports users in a competitive environment as defined in claim 9 wherein the specific contents of step 4 are:
processing the candidate positions popped up by the HU large top heap one by one according to the step 3, and when the competition influence value upper bound of the candidate positions at the top of the heap is smaller than the threshold score T Or until the last element of the large top heap HU is popped up and the processing is completed, the k candidate positions in the small top heap hs are the optimal k candidate positions.
11. A method for selecting a facility address based on influencing a sports user in a competitive environment as claimed in claim 9 wherein if the cost effectiveness of k candidate locations is considered in step 4, the cost per unit competitive influence value is utilized to make further selections for any of the optimal candidate locations c i If the cost of building a service facility at this location is cost (c i ) Then c can be calculated i Cost-effectiveness ratio, i.e. competitiveness, of (c) And calculating the competitiveness of the k optimal candidate positions based on the calculated competition force, and sequencing the k optimal candidate positions, wherein the maximum competitiveness is the optimal candidate position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010819078.1A CN112308597B (en) | 2020-08-14 | 2020-08-14 | Method for selecting facility address according to influence of sports user in competition environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010819078.1A CN112308597B (en) | 2020-08-14 | 2020-08-14 | Method for selecting facility address according to influence of sports user in competition environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112308597A CN112308597A (en) | 2021-02-02 |
| CN112308597B true CN112308597B (en) | 2023-07-18 |
Family
ID=74483193
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010819078.1A Active CN112308597B (en) | 2020-08-14 | 2020-08-14 | Method for selecting facility address according to influence of sports user in competition environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112308597B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103324748A (en) * | 2013-07-04 | 2013-09-25 | 上海交通大学 | Dynamic monitoring method and dynamic monitoring system for searching optimal competitive location |
| CN104346444A (en) * | 2014-10-23 | 2015-02-11 | 浙江大学 | Optimum site selection method based on road network reverse spatial keyword query |
| CN106651019A (en) * | 2016-12-14 | 2017-05-10 | 中国联合网络通信有限公司吉林省分公司 | Unicom operator channel site selection method |
| CN107423299A (en) * | 2016-05-24 | 2017-12-01 | 华为技术有限公司 | A kind of site selecting method and device |
| CN110675177A (en) * | 2018-07-03 | 2020-01-10 | 百度在线网络技术（北京）有限公司 | Store site selection method and device |
| CN111242694A (en) * | 2020-01-18 | 2020-06-05 | 曾海平 | Movable mall |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070192170A1 (en) * | 2004-02-14 | 2007-08-16 | Cristol Steven M | System and method for optimizing product development portfolios and integrating product strategy with brand strategy |
-
2020
- 2020-08-14 CN CN202010819078.1A patent/CN112308597B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103324748A (en) * | 2013-07-04 | 2013-09-25 | 上海交通大学 | Dynamic monitoring method and dynamic monitoring system for searching optimal competitive location |
| CN104346444A (en) * | 2014-10-23 | 2015-02-11 | 浙江大学 | Optimum site selection method based on road network reverse spatial keyword query |
| CN107423299A (en) * | 2016-05-24 | 2017-12-01 | 华为技术有限公司 | A kind of site selecting method and device |
| CN106651019A (en) * | 2016-12-14 | 2017-05-10 | 中国联合网络通信有限公司吉林省分公司 | Unicom operator channel site selection method |
| CN110675177A (en) * | 2018-07-03 | 2020-01-10 | 百度在线网络技术（北京）有限公司 | Store site selection method and device |
| CN111242694A (en) * | 2020-01-18 | 2020-06-05 | 曾海平 | Movable mall |
Non-Patent Citations (6)
| Title |
|---|
| 动态竞争环境下基于客户价值的物流终端设施优化模型;韩霜;周芳娟;谭智华;程南该;;公路交通科技(第11期);全文 * |
| 基于RkNN的空间位置影响力评价与查询算法;许景科;孙焕良;王永会;宋晓宇;;计算机研究与发展(第S3期);全文 * |
| 层次分析法在连锁便利店选址评价中的应用研究;王海燕;;洛阳师范学院学报(第04期);全文 * |
| 模糊决策法在大型超市选址合理性评价中的应用;李伟才;赵丽琴;肖丽华;;江西科学(第02期);全文 * |
| 超市选址相关指标分析;王英;;科技风(第14期);全文 * |
| 饭店选址影响因素的计量实证;穆学清;;区域治理(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112308597A (en) | 2021-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ke et al. | Lightgbm: A highly efficient gradient boosting decision tree | |
| US20210158164A1 (en) | Finding k extreme values in constant processing time | |
| CN110232403B (en) | Label prediction method and device, electronic equipment and medium | |
| CN103902597B (en) | The method and apparatus for determining relevance of searches classification corresponding to target keyword | |
| CN103888541B (en) | Method and system for discovering cells fused with topology potential and spectral clustering | |
| CN111382283B (en) | Resource category label labeling method and device, computer equipment and storage medium | |
| CN108520284A (en) | A kind of improved spectral clustering and parallel method | |
| CN108711103A (en) | Personal loan repays Risk Forecast Method, device, computer equipment and medium | |
| Jing et al. | AutoRSISC: Automatic design of neural architecture for remote sensing image scene classification | |
| US20210019635A1 (en) | Group specific decision tree | |
| CN114254615A (en) | Volume assembling method and device, electronic equipment and storage medium | |
| CN103218419B (en) | Web tab clustering method and system | |
| CN116304518A (en) | Heterogeneous graph convolution neural network model construction method and system for information recommendation | |
| Liu et al. | Detecting community structure for undirected big graphs based on random walks | |
| JP2018517963A (en) | Information processing apparatus, information processing method, and program | |
| Zhang et al. | Coresets remembered and items forgotten: submodular maximization with deletions | |
| CN107291894A (en) | A kind of probability matrix decomposition model for merging similitude and common scoring item quantity | |
| CN112308597B (en) | Method for selecting facility address according to influence of sports user in competition environment | |
| JP7498393B2 (en) | Information processing device, information processing method, program, and information processing system | |
| CN118332008A (en) | Answer screening method, device, computer equipment and storage medium | |
| CN116108076B (en) | Materialized view query method, materialized view query system, materialized view query equipment and storage medium | |
| CN108920492B (en) | Webpage classification method, system, terminal and storage medium | |
| Yan et al. | FLPA: A fast label propagation algorithm for detecting overlapping community structure | |
| CN110796198A (en) | High-dimensional feature screening method based on hybrid ant colony optimization algorithm | |
| CN110222055A (en) | The single-wheel core value maintaining method of multiple edge update under a kind of Dynamic Graph |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |