KR20200128236A - Automatic widget applying apparatus and method based on Machine-Learning - Google Patents

Abstract

Provided are a machine learning based automatic widget application device and a method thereof. The automatic widget application device comprises: a data collection part that collects data by performing a query from a data collection target server; a data analysis part that automatically determines a widget to be applied to the collected data by applying the collected data to a widget-applied learning model generated through machine learning (ML); a data binding part that automatically binds the collected data to the widget determined by the data analysis part; and a widget processing part that automatically visualizes the data-bound widget in a displayable form.

Description

Automatic widget applying apparatus and method based on Machine-Learning

The present invention relates to a machine learning-based automatic widget application device and method, and more particularly, a machine learning-based automatic widget application device that can automatically determine and apply a widget to be applied to collected data based on machine learning. And a method.

In the past, when creating a widget that visually shows the data collected from the data collection target server, the user manually creates the widget and directly selects the data to be bound to the widget.

One of the methods used to visually display the data collected from the data collection target server is a widget. When data is collected, the user manually creates one by one widget to show the data and places it on the screen, and binds the collected data directly to the widget. Therefore, whenever data is collected from a new data collection target server, the user must manually create a widget and repeat the operation of binding the data to the widget.

Korean Patent Publication No. 10-2010-0134495 (published on December 23, 2010)

In order to solve the above-described problem, the technical problem to be achieved by the present invention is to provide a machine learning-based automatic widget application apparatus and method capable of automatically applying a widget suitable for the data attribute by analyzing data based on machine learning. Have.

The problem of the present invention is not limited to those mentioned above, and other problems that are not mentioned will be clearly understood by those skilled in the art from the following description.

As a means for solving the above-described technical problem, according to an embodiment of the present invention, a machine learning-based automatic widget application apparatus includes: a data collection unit for collecting data by performing a query from a data collection target server; A data analysis unit for automatically determining a widget to be applied to the collected data by applying the collected data to a widget-applied learning model generated through machine learning (ML); A data binding unit for automatically binding the collected data to the widget determined by the data analysis unit; And a widget processing unit for visually processing the automatically data-bound widget in a displayable form.

The data analysis unit analyzes the properties of the collected data by inputting the collected data into the widget-applied learning model, and calculates a similarity rate by comparing the properties of the analyzed data with the properties of a plurality of preset widgets. And, among the calculated similarity rates, one of widgets having a similarity rate greater than or equal to a reference similarity rate is determined as a widget to be applied to the collected data.

The data analysis unit, for the collected data, analyzes the arrangement pattern of numerical values and strings, analyzes the arrangement and size of special characters of the numerical values and strings, analyzes the use and use of data field names, analyzes the number of columns and rows, and size of the timeline. And at least one of type analysis is performed to analyze the attribute of the data, and compare the analysis result with the attributes of the plurality of previously generated widgets.

When the component of the widget to which the data is bound is changed by a user request, a behavior pattern application unit that updates the change pattern of the component of the widget to which the data is bound and the property of the data bound to the widget to a machine learning DB; Further comprising, wherein the data analysis unit trains the widget-applied learning model using the change pattern stored in the machine learning DB and the property of the data bound to the widget, and then, the property of the data stored in the machine learning DB When new data similar to that is collected, the component automatically applies the changed widget to the new data.

On the other hand, according to another embodiment of the present invention, a machine learning-based automatic widget application method includes the steps of: (A) an automatic widget application device, collecting data by performing a query from a data collection target server; (B) The automatic widget application device automatically applies the widget to be applied to the collected data by applying the data collected in the step (A) to the learning model applying the widget generated through machine learning (ML). Judging; (C) automatically binding the collected data to the widget determined in step (B), by the automatic widget application device; And (D) visually processing, by the automatic widget application device, the widget to which data is automatically bound in the (C) step in a displayable form.

The step (B) includes: (B1) analyzing the properties of the collected data by inputting the collected data to the widget-applied learning model; (B2) calculating a similarity rate by comparing the properties of the analyzed data and properties of a plurality of widgets set in advance; And And (B3) determining one of the widgets having a similarity rate greater than or equal to a reference similarity rate among the calculated similarity rates as a widget to be applied to the collected data.

(E) When the automatic widget application device changes the component of the widget to which the data is bound by a user request, machine learning the change pattern of the component of the widget to which the data is bound and the properties of the data bound to the widget Updating the DB; (F) learning, by the automatic widget application device, the widget application learning model using the change pattern stored in the machine learning DB and the property of data bound to the widget; And (G) when the automatic widget application device collects new data similar to the property of the data stored in the machine learning DB, automatically applying the widget whose component is changed to the new data.

According to the present invention, it is possible to maximize the use of resources by increasing recyclability of data having the same pattern, thereby increasing the expected cost.

Further, according to the present invention, various data may be provided by performing various queries through a user-centered simple UI interface.

In addition, according to the present invention, it is possible to improve the quality of the visualization data binding by automatically generating raw numeric data or character strings as a widget that can improve readability and readability and then binding them.

In addition, according to the present invention, it is possible to more easily access main information by extracting only necessary information through machine learning from various and vast amounts of data, and reducing unnecessary searches.

In addition, according to the present invention, it is possible to extract data satisfying conditions in various cases with only a simple input and automatically express it through a widget.

The effects of the present invention are not limited to those mentioned above, and other effects that are not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a block diagram showing a machine learning-based automatic widget application apparatus according to an embodiment of the present invention;
2 is a diagram illustrating an example of a plurality of widgets generated for each data type;
3 is an exemplary view of a query screen,
4 is a flowchart schematically showing a machine learning-based automatic widget application method of an automatic widget application device according to an embodiment of the present invention;
Figure 5 is a flow chart showing in detail step S420 of Figure 4, and,
6 is a diagram schematically illustrating an operation of the apparatus for applying an automatic widget described with reference to FIGS. 1 to 5.

Prior to describing specific details for the implementation of the present invention, terms or words used in the specification and claims may be appropriately defined by the inventor in order to describe his or her own invention in the best way. Based on the principle that there is, it should be interpreted as a meaning and concept corresponding to the technical matters of the present invention.

Terms such as "comprise", "comprise", and "have" described in the present specification mean that the corresponding component may be included unless otherwise specified, so other components are not excluded. It means that it can contain more elements.

In addition, terms such as "... unit", "... group", and "module" described in this specification mean units that process at least one function or operation, which can be implemented by hardware or software or a combination of hardware and software. I can. In addition, articles such as "one", "one" and "the" are meant to include both the singular and the plural in the context describing the present invention, unless otherwise indicated in the specification or clearly contradicted by the context. Can be used.

In addition, if an element, component, device, or system is stated to include a program or a component made of software, the element, component, device, or system is executed by the program or software, even if there is no explicit mention. Or, it should be understood to include hardware (eg, memory, CPU, etc.) or other programs or software (eg, a driver required to drive an operating system or hardware) required to operate.

In addition, it should be understood that an element (or component) may be implemented in software, hardware, or any form of software and hardware, unless otherwise specified in the implementation of any element (or component).

The above objects, other objects, features, and advantages of the present invention will be easily understood through the following preferred embodiments related to the accompanying drawings. However, the present invention is not limited to the embodiments described herein and may be embodied in other forms. Rather, the embodiments introduced herein are provided so that the disclosed content may be thorough and complete, and the spirit of the present invention may be sufficiently conveyed to those skilled in the art.

Hereinafter, the present invention will be described in detail with reference to the drawings.

1 is a block diagram showing an apparatus 100 for applying an automatic widget based on machine learning according to an embodiment of the present invention.

The machine learning-based automatic widget application apparatus 100 according to the embodiment of the present invention shown in FIG. 1 collects data and then analyzes the pattern of the data stored in the machine learning DB 140 to determine the properties of the collected data. Appropriate widgets are automatically applied, data utilization patterns are stored in the machine learning DB 140 in real time, and rulesets can be applied to each widget. In addition, the automatic widget application device 100 may bind the collected data to the automatically applied widget and provide it to the user. The automatic widget application device 100 may be a dashboard builder or an electronic device for the operation of the dashboard builder.

Referring to FIG. 1, a machine learning-based automatic widget application apparatus 100 according to an embodiment of the present invention includes a user interface unit 110, a display unit 120, a communication interface unit 130, and a machine learning DB 140. , May include a memory 150 and a processor 160.

The user interface unit 110 provides an interfacing path between the user and the automatic widget application device 100. For example, a user may manipulate the user interface unit 110 to generate a widget for each resource type of various data.

2 is a diagram illustrating an example of a plurality of widgets generated for each data type.

Referring to FIG. 2, when a user sets a component or property of a widget differently or similarly for each resource type of data, the widget generator 161 may generate a plurality of widgets by referring to the component or property set by the user. have. The components of the widget are elements to visualize data such as bar chart, bubble chart, line chart, timeline chart, grid, image, heat map, alert, field map, symbol map, flow map, table, text, etc. to be. The properties of the widget may be different for each widget or for each component, such as an arrangement pattern of values and strings set when creating the widget, data field names, usage, columns and rows, units of charts, size and type of timeline, and so on.

The display unit 120 includes new layers used by the user interface unit 110 to create a widget, components, widgets generated by the widget creation unit 161, and data by the widget processing unit 165 to be described later. Displays various information to be provided to the user, such as a widget that has been bound and processed.

The communication interface unit 130 may be connected to the data collection target server 10 to receive data generated by the data collection target server 10. For example, the communication interface unit 130 attempts a TCP/IP connection to the data collection target server 10 through the data adapter, and when the data adapter connection is completed, the communication interface unit 130 may receive data from the data collection target server 10. have.

The machine learning DB 140 applies automatic widgets, such as initial data received from the data collection target server 10, data collected by executing a query suitable for each content, and a plurality of widgets generated by the widget generator 161 It is possible to store and update a plurality of data, information, etc. generated in the device 100.

The memory 150 may include volatile memory and/or nonvolatile memory. In the memory 150, for example, in order to implement and/or provide an operation and a function provided by the electronic device 200, commands or data related to the components 210 to 260, one or more programs, and/or Software, operating system, etc. may be stored.

Programs stored in the memory 150 include a widget creation program for generating a widget, a machine learning program for generating a widget-applied learning model through machine learning (ML), and a widget based on the type of collected data. A widget application learning model generated by machine learning in advance in order to automatically apply a widget, and a widget application program for binding data collected to a widget automatically applied through the widget application learning model may be included.

The processor 160 controls the overall operation of the automatic widget application device 100 by executing one or more programs stored in the memory 150.

For example, the processor 160 may generate a widget by executing a widget generation program stored in the memory 150. To this end, the processor 160 may include a widget generator 161. The widget creation unit 161 generates widget creation screens and displays them on the display unit 120, and a plurality of widget creation screens displayed on the display unit 120, the properties of the widget received through the user interface unit 110 A number of widgets may be created with reference to (or widget creation conditions), and the created widgets may be stored in the machine learning DB 140. Thereafter, when the user creates a new layer on the visualization desktop and places the created widgets on each layer, selects a data adapter and a data collection target server, and selects the automatic widget application menu, the processor 160 returns the collected data. The optimal widget can be automatically applied according to the property (or type) of.

The properties of the widget received through the user interface unit 110 may be used to generate a widget application learning model, and the widget application program to analyze the type of data collected through the widget application learning model. The properties of the widget are the layout pattern of numbers and strings set to implement the dashboard, the arrangement and size of special characters of the numbers and strings, the purpose and use of data field names set to implement the dashboard, and implement the dashboard. It includes the number of columns and rows set to do so, the size and type of the timeline, and all properties set for implementing the dashboard.

In addition, the processor 160 executes the machine learning program stored in the memory 150 to machine-learn properties set in the widgets stored in the machine learning DB 140 and the widgets mapped to the properties to generate a widget-applied learning model. can do.

In addition, the processor 160 executes the widget application program and applies the collected data to the widget application learning model generated through machine learning, automatically determines the widget to be applied to the collected data, and applies the collected data to the widget. Can be bound. Further, after the widget is automatically applied to the collected data, if the property of the initial widget is changed, the processor 160 may update the widget application learning model based on the pattern in which the property is changed.

To this end, the processor 160 includes a data collection unit 162, a data analysis unit 163, a data binding unit 164, a widget processing unit 165, a behavior pattern application unit 166, and a ruleset definition unit 167. It may contain more.

The data collection unit 162 attempts a TCP/IP connection to connect to the data collection target server 10 through the data adapter, and when the data adapter connection is completed, performs a data query to receive the data from the data collection target server 10. Collect data. The user sets a query, that is, a search condition, through the query screen as shown in FIG. 3, and the data collection unit 162 collects data corresponding to the set query and stores it in the machine learning DB 140.

The data analysis unit 163 applies the data collected by the data collection unit 162 to the widget application learning model generated through machine learning, and analyzes a number of conditions for automatically applying the widget to the type of the collected data. Is determined, and a widget to be applied to the data may be automatically determined based on the determined data type.

In detail, the data analysis unit 163 inputs the data collected by the data collection unit 162 into a widget-applied learning model, analyzes various properties of the collected data, determines the type of data, and determines the type of the analyzed data. The similarity rate is calculated by comparing the properties of a plurality of preset widgets.

The data analysis unit 163 or the learning model applying a widget, for the collected data, analyzes the arrangement pattern of numbers and strings, analyzes the arrangement and size of special characters of numbers and strings, analyzes the use and utilization of data field names, the number of columns and rows Analyze various properties of data by performing at least one of a number of analysis methods including analysis and timeline size and type analysis, and calculate the similarity rate by comparing the analysis result with the properties of a plurality of previously generated widgets can do. This means that properties such as the arrangement pattern of multiple numbers and strings, the arrangement and size of special characters of numbers and strings, the purpose and use of data field names, the number of columns and rows, and the size and type of the timeline vary depending on the type of collected data. This is because it has a value and is set differently according to the properties (or roles) of each of the para-generated widgets.

In addition, the data analysis unit 163 may determine a widget having a maximum similarity rate among the calculated similarity rates as a widget to be applied to the collected data. Alternatively, the data analysis unit 163 may determine one of widgets having a similarity rate greater than or equal to the reference similarity rate among the calculated similarity rates as a widget to be applied to the collected data. For example, when there are three widgets with a reference similarity rate of 95% and a calculated similarity rate of 95% or more, the data analysis unit 163 applies the widget with the highest similarity rate among the three widgets to the collected data. It can be determined as a widget to do.

The data binding unit 164 may automatically bind the collected data to the widget determined by the data analysis unit 163.

The widget processing unit 165 visually processes the widget to which data is automatically bound by the data binding unit 164 in a displayable form.

When the component of the widget to which data is bound is changed by a user request, the behavior pattern application unit 166 transmits the change pattern of the component of the widget to which data is bound and the properties of the data bound to the widget to the machine learning DB 140 Can be updated.

The data analysis unit 163 may learn a widget application learning model by using a change pattern of a widget updated in the machine learning DB 140 and a property of data bound to the widget to change a component of the widget according to a user request. That is, when the same change pattern is input for the same widget through the user interface unit 110 as many times as a set number of times, the behavior pattern application unit 166 may change the component of the initially created widget as requested by the user. Thereafter, when new data similar to the attribute of the data stored in the machine learning DB 140 is collected, the data analysis unit 163 may automatically apply a widget whose attribute or component has been changed to the new data.

For example, if the target widget applied to the collected data is a map widget that shows only Korea, the user requests to change the pattern to show all of Asia, including Korea, and if the change request is repeated a set number of times, the behavior pattern The application unit 166 changes the component of the target widget from Korea to an Asia map. Thereafter, if the property of the newly collected data is similar to the property of the target widget, the collected data is bound to the target widget, thereby providing a widget showing an Asia map.

The rule set definition unit 167 may collectively define a rule set in the automatically applied widget according to the properties of the collected data. When a user inputs a rule set to be applied to a widget identically or differently for each widget through the user interface unit 110, the rule set definition unit 167 processes the input rule set for each widget to be applied to each widget. Subsequently, when data is collected from the data adapter, an alarm or effect may be expressed by a rule set previously applied to the corresponding widget. For example, the ruleset definition unit 167 defines a rule set in the widget that causes blue flicker to occur with respect to data on students of 5th grade or higher of elementary school among the collected data. Each time the widget flashes blue.

4 is a flowchart schematically illustrating a machine learning-based automatic widget application method of an automatic widget application apparatus according to an embodiment of the present invention, and FIG. 5 is a detailed flowchart illustrating step S420 of FIG. 4.

Referring to FIG. 4, the automatic widget application device 100 collects data by performing a query from the data collection target server 10 and stores it in a machine learning DB (S410).

The automatic widget application device 100 automatically determines a widget to be applied to the collected data by applying the collected data to a widget-applied learning model generated through machine learning (ML) (S420).

Referring to FIG. 5, the automatic widget application apparatus 100 inputs the collected data into the widget application learning model and analyzes the properties of the collected data, that is, the type of data (S422). Step S422, for the collected data, analyzes the arrangement pattern of numbers and strings, analyzes the arrangement and size of special characters of numbers and strings, analyzes the use and use of data field names, analyzes the number of columns and rows, and analyzes the size and type of the timeline. You can perform at least one of to analyze the properties of the data.

The automatic widget application apparatus 100 calculates a similarity rate by comparing the properties of the analyzed data with the properties of a plurality of widgets set in advance (S424).

The automatic widget application apparatus 100 determines one of the widgets having a similarity rate equal to or greater than the reference similarity rate among the calculated similarity rates (eg, a widget having a maximum similarity rate) as a widget to be applied to the collected data (S426). ).

Referring to FIG. 4 again, the automatic widget application device 100 automatically binds the data collected in step S410 to the widget determined in step S420 (S430).

The automatic widget application device 100 automatically visualizes a widget to which data is bound in a displayable form and displays it on the screen (S440).

Thereafter, when the component of the widget displayed on the screen, that is, the component of the widget to which data is bound, is changed at the request of the user, the automatic widget application device 100 binds to the widget and the change pattern of the component of the widget to which data is bound The properties of the obtained data are updated in the machine learning DB 140 (S450).

The automatic widget application device 100 trains the widget application learning model using the change pattern stored in the machine learning DB 140 and the property of data bound to the widget (S460). Step S460 may be performed when step S450 is repeated a predetermined number of times.

When new data similar to the property of the data stored in the machine learning DB 140 is collected, the automatic widget applying device 100 automatically applies the widget whose component has been changed to the new data (S470). Accordingly, even if new data similar or identical to the property of the data collected in step S422 is collected, the automatic widget application apparatus 100 applies the widget whose component is changed in step S450, not the widget applied in step S420.

The automatic widget application apparatus 100 defines and stores a rule set such as a data threshold value or attribute values of various components in the automatic widget applied to the collected data type (S480).

6 schematically illustrates the operation of the automatic widget application device 100 described with reference to FIGS. 1 to 5, and the automatic widget application device 100 performs data pattern analysis of the machine learning DB 140 after data collection. It is possible to automatically determine a widget that fits the properties of the collected data, store the behavior pattern of the data (i.e., the widget pattern changed by the user) in the machine learning DB 140 in real time, and then apply the rule set at once. .

In particular, when data for each query is collected by performing another query on the data collection target server 10, a data type may be determined for each data collected for each query, and a widget may be automatically applied individually. That is, since the type of data to be collected varies according to the query (data types 1 to n), different data may be collected for each query even for data generated in the data collection target server 10. Accordingly, a plurality of different data may be collected for one data collection target server 10 and different widgets may be applied.

In addition, in the data behavior pattern analysis, when a behavior pattern change occurs that automatically changes the component or attribute of the widget to which data is assigned, the attribute of the data and the changed pattern applied to the corresponding widget are updated in the machine learning DB 140, and later When similar data is collected or bound, it can be reassigned to the changed widget.

According to the embodiment of the present invention described above, a widget for each data type is automatically generated through data properties and behavior patterns, and the properties of the existing widgets are automatically modified according to a change in the use pattern of data. Data widgets based on various components such as charts, grids, images, alerts, etc. can be automatically changed to grid or line chart components and applied by changing the usage pattern of data after creating an initial pie component.

100: automatic widget application device 110: user interface unit
120: display unit 130: communication interface unit
140: machine learning DB 150: memory
160: processor

Claims (8)

A data collection unit for collecting data by performing a query from a data collection target server;
A data analysis unit for automatically determining a widget to be applied to the collected data by applying the collected data to a widget-applied learning model generated through machine learning (ML);
A data binding unit for automatically binding the collected data to the widget determined by the data analysis unit; And
And a widget processing unit for visually processing the automatically data-bound widget in a displayable form. The method of claim 1,
The data analysis unit,
The collected data is input into the widget-applied learning model to analyze the properties of the collected data, compare the properties of the analyzed data with the properties of a plurality of preset widgets to calculate a similarity rate, and the calculated A machine learning-based automatic widget application device, characterized in that, among the similarity rates, one of widgets having a similarity rate greater than or equal to a reference similarity rate is determined as a widget to be applied to the collected data. The method of claim 2,
The data analysis unit,
For the collected data, at least one of analysis of the arrangement pattern of numerical values and strings, analysis of the arrangement and size of special characters of the numerical values and strings, analysis of data field names use and utilization, analysis of the number of columns and rows, and analysis of the size and type of the timeline. A machine learning-based automatic widget application device, characterized in that performing one to analyze the properties of the data, and comparing the analysis result with the properties of the plurality of previously generated widgets. The method of claim 1,
When the component of the widget to which the data is bound is changed by a user request, a behavior pattern application unit that updates the change pattern of the component of the widget to which the data is bound and the property of the data bound to the widget to a machine learning DB; Including more,
The data analysis unit,
The widget-applied learning model is trained using the change pattern stored in the machine learning DB and the property of the data bound to the widget, and then, when new data similar to the property of the data stored in the machine learning DB is collected, the Machine learning-based automatic widget application device, characterized in that automatically applying a widget whose component has been changed to the new data. (A) the automatic widget application device, collecting data by performing a query from a data collection target server;
(B) The automatic widget application device automatically applies the widget to be applied to the collected data by applying the data collected in the step (A) to the learning model applying the widget generated through machine learning (ML). Judging;
(C) automatically binding the collected data to the widget determined in step (B), by the automatic widget applying device; And
(D) the automatic widget applying device visually processing the widget to which the data is automatically bound in the (C) step in a displayable form; and a machine learning-based automatic widget application method comprising: The method of claim 5,
The (B) step,
(B1) analyzing the properties of the collected data by inputting the collected data into the widget-applied learning model;
(B2) calculating a similarity rate by comparing the properties of the analyzed data and properties of a plurality of preset widgets; And
(B3) determining one of the widgets having a similarity rate greater than or equal to a reference similarity rate among the calculated similarity rates as a widget to be applied to the collected data; a machine learning-based automatic widget application method comprising . The method of claim 6,
The step (B1),
For the collected data, at least one of analysis of the arrangement pattern of numerical values and strings, analysis of the arrangement and size of special characters of the numerical values and strings, analysis of data field names use and utilization, analysis of the number of columns and rows, and analysis of the size and type of the timeline. Machine learning-based automatic widget application method, characterized in that performing one to analyze the properties of the data. The method of claim 5,
(E) When the automatic widget application device changes the component of the widget to which the data is bound by a user request, machine learning the change pattern of the component of the widget to which the data is bound and the properties of the data bound to the widget Updating the DB;
(F) learning, by the automatic widget application device, the widget application learning model using the change pattern stored in the machine learning DB and the property of data bound to the widget; And
(G) when the automatic widget application apparatus collects new data similar to the property of the data stored in the machine learning DB, automatically applying the widget whose component has been changed to the new data; characterized in that it further comprises How to apply automatic widgets based on machine learning.

Images