JP2019070913A - Schedule information analysis system and program - Google Patents

Abstract

To provide a method which analyzes how to use, distribution and tendency of labour hours from information of a scheduler, reconstitutes input information and visualizes the contents by means of outputting them as dashboard information.SOLUTION: A schedule information analysis system 1 which generates prescribed estimated schedule information based on schedule information expressing user's planned contents and planned dates inputted into schedule information input means (scheduler) 31 via user terminals 30, is constituted of schedule information acceptance means 21 to receive schedule information inputs, estimated schedule information generating means 22 to classify schedule information into prescribed group units by means of machine learning title sentences included in the schedule information and to generate prescribed estimated schedule information and a dashboard display means 23 to generate the estimated schedule information as a visualized prescribed dashboard information.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a technique for analyzing schedule information input by workers, and in particular, to generate visualized information by sorting the usage, distribution, and trends of working hours of individual workers and the entire organization. Systems and programs.

  Individuals' attendance and leaving work and work schedules are sometimes managed using software called a scheduler, and each worker enters and registers their own schedule etc. in the scheduler to identify the person and affiliation thereof. Can be shared by

  Here, as a technique related to this type of scheduler, related information such as "restaurant ○ ○" is displayed / displayed with respect to schedule information input to the scheduler, for example, "meal" information. A system is disclosed.

Japanese Patent Application Publication No. 2007-065911

  However, the technology disclosed in Patent Document 1 displays additional information on the premise that input information is correct, and the input information is a unique expression, or lack of information or mutual contradiction occurs. If this is the case, additional information based on that will also be useless.

  The present invention aims to provide a schedule information analysis system and program for correctly analyzing and visualizing input information in order to solve the above problems, and to contribute to the reform of working manner.

  In order to achieve the above object, a schedule information analysis system of the present invention is a schedule information analysis system for generating predetermined estimated schedule information based on schedule information indicating schedule contents and scheduled date and time of a user, the schedule information The schedule information receiving means for receiving the input of the above and the subject sentence included in the schedule information by machine learning to classify the schedule information into a predetermined group unit and generate the estimated schedule information to generate the predetermined estimated schedule information And means.

Furthermore, the present invention can be configured as a program executed by the schedule information analysis system according to the present invention as described above.
Furthermore, the present invention can also be configured as a method that can be implemented by the schedule information analysis system and program according to the present invention as described above.

According to the invention, rather than just making the scheduler more visible and displaying relevant information,
By sorting subject texts input to the scheduler using artificial intelligence, it is reconstructed as estimated schedule information that analyzes usage, distribution, trends, etc. of working hours, and the content can be recognized and grasped at a glance. Can be generated as the generated dashboard information.
As a result, based on the generated information, it becomes possible to realize the actual status of working hours and working conditions, and to improve the productivity of individuals and the entire workplace and reform of working styles.

It is a functional block diagram showing composition of each device in a schedule information analysis system concerning one embodiment of the present invention. It is a flowchart which shows the processing operation in the system. It is explanatory drawing which shows an example of the schedule information output from the user terminal of the system. It is explanatory drawing which shows an example of the procedure of machine learning of the schedule information in the same system. It is explanatory drawing which shows an example in the case of labeling the result of the machine learning of the schedule information in the same system. It is explanatory drawing which shows an example of the entrance / exit data in the system. It is an explanatory view showing an example of position information used for analysis of schedule information in the same system. It is explanatory drawing which shows an example of the logic for producing | generating presumed schedule information from the schedule information which overlaps in the same system. It is explanatory drawing which shows an example of the dashboard information produced | generated in the system. It is explanatory drawing of the display content of the dashboard information shown in FIG. It is explanatory drawing which shows another example of the dashboard information produced | generated in the system.

Hereinafter, embodiments of a schedule information analysis system and program according to the present invention will be described with reference to the drawings.
Here, a schedule information analysis system (hereinafter, simply referred to as "the present system") of the present invention shown below is realized by processes, means, and functions executed by a computer by instructions of a program (software). The program can send an instruction to each component of the computer to perform predetermined processing or functions according to the present invention described below. That is, each process, means, and function in the present invention are realized by specific means in which a program and a computer cooperate.

  Note that all or part of the program is provided by, for example, a magnetic disk, an optical disk, a semiconductor memory, or any other computer-readable recording medium, and the program read from the recording medium is installed in a computer and executed. Ru. The program can also be loaded and executed directly on a computer through a communication line without using a recording medium. Further, the present system can be configured by a single information processing apparatus (for example, one personal computer) or can be configured by a plurality of information processing apparatuses (for example, a plurality of server computer groups).

[System configuration]
As shown in FIG. 1, the present system 1 is configured to include a schedule information management server 10, an administrator device 20, and one or more user terminals 30 (30a to 30n), and the user's input Predetermined estimated schedule information is generated on the basis of schedule information indicating schedule contents and scheduled dates and times of the user, which are input by the user terminal 30 and other devices / terminals in response to the operation.
The schedule information management server 10, the administrator device 20, and the user terminal 30, which constitute the present system 1, are connected via a predetermined network 40 such as the Internet, LAN, WAN, etc., for example, and can mutually communicate with each other. ing.

[Schedule information management server 10]
The schedule information management server 10 is an information processing apparatus that manages a scheduler (schedule software) to which schedule information is input via the user terminal 30 or the like, and, for example, on one or more server computers or cloud computing services It can be configured by a server system or the like consisting of one or more virtual servers constructed.
For example, the schedule information management server 10 can be configured by, for example, a scheduler server on the Internet (an Express server that manages data of Microsoft “Outlook” (registered trademark) that is a scheduler) or the like.

Specifically, as shown in FIG. 1, the schedule information management server 10 includes a schedule information generation unit 11.
The schedule information generation unit 11 takes in the schedule information input to the scheduler via the user terminal 30 or the like, and converts / generates it as schedule data in a predetermined data format (for example, “CSV” or “XLS” format), The schedule data is spouted and taken into the administrator device 20.

In the present embodiment, the schedule information management server 10 takes in the information of the scheduler and discharges it to the administrator device 20 as schedule data of a predetermined format, but this function (schedule information generation means 11) is managed It can also be provided on the person device 20 side or the user terminal 30 side. For example, the schedule information generation unit 11 can be configured by a module, a dedicated macro or the like mounted on the administrator device 20. Further, the schedule information is output in a predetermined data format (for example, “XLS” or “CSV” format) using a data output function provided as a scheduler function, and the schedule information is output to the administrator device 20. It can also be directly taken in without the intervention of the management server 10.
In that case, the administrator device 20 or the user terminal 30 is provided with the schedule information generation means 11, and the present system 1 is configured as a system not provided with the schedule information management server 10.

[Administrator device 20]
The administrator device 20 can be configured by, for example, one or more server computers, a server system including one or more virtual servers built on a cloud computing service, or the like.
The administrator device 20 is provided with an OS (Operating System), a DBMS (DataBase Management System), and the like (not shown), and is operated as a server computer.

In addition, software operating on middleware such as a web server program is implemented in the administrator device 20.
Further, the administrator device 20 is provided with a storage unit implemented as a database or the like, and is provided with a storage unit that constructs a database and acquires and accumulates predetermined information necessary for the operation of the present system 1.
The storage means stores, as various information resources, schedule information input by the user terminal 30 described later and estimated schedule information / dashboard information generated based thereon, and the manager according to the present embodiment. The predetermined information required for the device 20 is stored, and the necessary information is read as needed along with the operation of the administrator device 20, and stored / updated.

And, as shown in FIG. 1, specifically, the manager device 20 as described above is configured to function as the schedule information reception means 21, the estimated schedule information generation means 22, and the dashboard display means 23. There is.
The schedule information receiving unit 21 is a unit that receives an input of schedule information (schedule data) input to the scheduler, which is output through the schedule information management server 10.
The acceptance of the schedule information can be downloaded from the above-described schedule information management server 10 and the schedule information can be fetched, or can be directly fetched from the user terminal 30 which has input the schedule information.

  The estimated schedule information generating means 22 is a means for classifying the schedule information into predetermined group units and generating estimated schedule information by machine learning the schedule information inputted and accepted by the schedule information accepting means 21. Estimated schedule information is generated by sorting the schedule information by artificial intelligence. In particular, in the present embodiment, the estimated schedule information generating means generates estimated schedule information by causing artificial intelligence to learn and sort text information constituting a sentence indicating “subject” included in the schedule information.

Specifically, the estimated schedule information generation unit 22 according to the present embodiment is (1) schedule information input to and registered in the scheduler via the user terminal 30 or the like as input information to be analyzed (see FIG. 3). Refer to), (2) time information (see Fig. 6) indicating the time of attendance and leaving (entry / exit) of the user corresponding to the schedule information, (3) attribute information including the identification information of the user corresponding to the schedule information Estimated schedule information is generated based on four types of information: (see FIG. 6) and (4) position information of a predetermined place including the work place of the user corresponding to the schedule information (see FIG. 7).
These input information are stored in the manager device 20, the manager device, and the like through the schedule information management server 10, the user terminal 30, the entrance / exit system of the company where the user works, the employee management system, etc. 20 are stored in accessible storage means.

Then, based on such input information, the estimated schedule information generation means 22 estimates the schedule information input to the scheduler and estimated information of blank time zones and dates not included in the schedule information. Generate as information.
Specifically, the estimated schedule information generation means 22 classifies, by AI, text information (see FIG. 3) indicating "subject" included in schedule information input to the scheduler via the user terminal 30 etc. as described above. By doing this, information (hereinafter referred to as "labeling information") classified into predetermined group units according to the contents of the schedule and the time is generated, and estimated schedule information is generated.

In addition, the estimated schedule information generation unit 22 generates estimated schedule information based on the above-described labeling information for the date and time that is not included in the schedule information input to the scheduler.
Here, as shown in FIG. 5, for example, as shown in FIG. 5, given labels “0: in-house meeting”, “1: outside meeting”, “5: in-house meeting”, from “subject” of scheduler (Outlook etc.) and entry / exit data. : It can be generated by temporarily creating an external form assigned “other than work”, and further assigning “2: move” and “3: work” labels in blanks, and estimated schedule information based on this labeling information Can be generated.

  Furthermore, the estimated schedule information generation means 22 is based on time information indicating the time of arrival and departure of the user corresponding to the labeling information and position information of a predetermined place including the work place of the user corresponding to the schedule information. If the labeling information includes multiple pieces of schedule information overlapping in time zones, any one piece of schedule information is extracted based on a predetermined rule to generate estimated schedule information (see FIG. 8).

The dashboard display means 23 is means for generating and displaying the estimated schedule information generated by the estimated schedule information generating means 22 as predetermined dashboard information (see FIGS. 9 to 11) which is visualized.
Specifically, the dashboard display means 23 displays the estimated schedule information generated by the estimated schedule information generation means 22 described above in a list in order of date and time with a predetermined period as a unit (division, affiliation, etc. ) As dashboard information visualized as unit image information (see FIG. 9).
The details of the process / operation for schedule information analysis executed by the respective units 21 to 23 of the manager device 20 as described above will be described later with reference to FIGS.

[User terminal 30]
The user terminal 30 is an information processing apparatus connected with the above-described administrator device 20 so as to be capable of data communication, and includes, for example, a PC, a tablet terminal, a smartphone, and the like.
Specifically, the user terminal 30 according to the present embodiment is configured to function as a schedule information input unit 31 and a schedule information output unit 32.

The schedule information input means 31 responds to the user's input operation to the user terminal 30, according to the user's schedule, for example, scheduled date and time such as business meeting, business trip, interview, delivery date, etc. For daily schedule, etc., “person's name” (user name etc.), “subject” (content, text indicating title etc.) or “place”, “start time”, “end time”, “owner”, “ It is a means to enter and register required attendees, "arbitrary attendees", "the person's reply status", etc. as text data.
Such schedule information input means 31 can be configured by software (application) called scheduler (schedule software).

  The scheduler is software that functions and is used as a "schedule book" that manages personal schedule information, and for example, functions such as a schedule, contact management, work management, memos, etc. are implemented, and an e-mail function is It is configured to be implemented or interlocked. As this type of scheduler, for example, Microsoft "Outlook" (registered trademark), IBM "Notes" (registered trademark), Cybozu "Galoon" (registered trademark), Google "Calendar", etc. are known.

In the present embodiment, the software that functions as such a scheduler can be configured as the schedule information input unit 31 by being installed in the user terminal 30.
The scheduler may also be disclosed and provided as an application that can be used in the form of an API (Application Programming Interface) via a network such as the Internet, and the API is called from the user terminal 30 via the scheduler application. Can be configured as the schedule information input means 31.

Then, the schedule information input unit 31 (scheduler) as described above receives the schedule information input into the schedule information management server 10, and the schedule information generation unit 11 converts the schedule information into a predetermined data format. Will be sent to
Specifically, of the schedule information inputted and stored in the schedule information input means 31, predetermined text data, for example, "person name", "subject", "place", "location time", "start time" of the schedule The text data indicating the end time, the "organizer", the "essential attendee", the "arbitrary attendee", the "individual reply status", etc. is extracted, and the administrator device 20 is extracted via the schedule information management server 10. It is captured.

The output of the schedule information to the administrator device 20 is performed by using the data output function provided as a function of the scheduler, in a predetermined data format (for example, “XLS” or “CSV” format) of the schedule information. It can also be output. In this case, the scheduler constitutes the schedule information generation means 11, and the schedule information is directly output from the user terminal 30 to the administrator device 20 via the network 40.
Then, the schedule information input through the user terminal 30 or the like as described above is input to the administrator device 20, and the estimated schedule information according to the present embodiment and the predetermined dashboard information visualizing the result are generated. Will be

[Operation]
Next, specific processes and operations (implementation of the schedule information analysis method) in the present system 1 as described above will be described with reference to FIGS.
First, as a premise of processing operation in the present system 1, "schedule data" (original data of scheduler) serving as schedule information to be input / stored through the user terminal 30 is accumulated as data to be analyzed.
In addition to the “schedule data”, “entry data”, “affiliation data of the measurement target employee”, and “location information such as office”, which are predetermined input information to the manager device 20, are accumulated.

"Schedule data"
As schedule data, the original data of the scheduler (schedule information input means 31) which each user inputs and manages is mentioned. For example, the data analysis in the present system is executed at the timing when data for all the measurement target persons are accumulated for a predetermined period (for example, the past six months to one year) for the employees of the corresponding company.
The original data of the scheduler, which becomes such schedule information, can be generated and output in one row of XLS or CSV format for each "one event" (schedule information output means 32). "Item", "Location", "Start Time", "End Time", "Host", "Required Attendee", "Any Attendee", "Person's Reply Status", etc. are included. (See Figure 3).

"Entry and exit data"
Entry and exit data includes, for example, data indicating the first entry time, last exit time, passage gate location at first entry, passage gate location at last exit, etc. of each employee of the corresponding company. (See FIG. 6).
These entry and exit data can be input and stored in the administrator device 20 from an entry and exit system of a company where the user works, a time recorder, or the like.
With regard to such entrance / exit data, data analysis in the present system is executed at a timing when a predetermined period (for example, the same period as the described schedule data) is accumulated for all the measurement subjects.

"Affiliation data of measurement target employee"
The affiliation data of the measurement target employee is, for all the measurement target employees, for example, a department name, a department name, a name, an alias (a unique ID identifying an employee, etc. such as an employee number or an e-mail) to which each employee of the corresponding company belongs. Data such as the left side of the address @ mark is included (see FIG. 6).
The affiliation data of this employee can be generated in the XLS or CSV format of one line for each “one person” in the same manner as the schedule data described above, and for each line, “employee number”, “name”, “ The respective items such as the belonging company and the belonging department are included (see FIG. 6).
If there is a movement of affiliation during the data collection / measurement period, or if the affiliation within the measurement period is different from the current affiliation, which affiliation is to be used is, for example, the length of affiliation period, analysis It can be determined according to the purpose.

"Position information such as office"
As position information of an office etc., there are latitude / longitude information such as an office registered as the "place" of the corresponding company, an in-house meeting room, an entrance / exit gate location etc. (see FIG. 7).
Similar to the schedule data and affiliation data described above, the location information of this office etc. can also be generated in XLS or CSV format of one line for each “one place”. Items such as "latitude" and "longitude" are included.
Here, for example, when there is one or two or more meeting rooms in one business office, the positional information and the entry / exit records about the “place” can be the same latitude and longitude.

For example, “Conference room 34: Latitude 34.2345, longitude 134.3453” is assumed to be the same position information as “Osaka east office: latitude 34.2345 · longitude 134.3453” for “conference room 34” in “Osaka east office” be able to.
Similarly, the “reception conference room A” in the “Tokyo head office” can be “Tokyo head office: latitude 35,5332, longitude 135.3323”.
In addition, although the above positional information (latitude and longitude information) may be provided from the beginning of the operation of this system, it can be made unnecessary at the beginning of the operation. For example, after operating the system, entering and analyzing schedule data and entry / exit data, add and enter necessary location information later depending on which "place" latitude / longitude information is required. Can.
In addition, latitude / longitude information can be set to any location using a map search site (for example, "Google Map" etc.) provided via the Internet or a geocoding service (for example, GEO Coding API for "Google" etc.) "Latitude and longitude information" can be easily obtained (see FIG. 7).

  As described above, in the present system 1, as input information to be analyzed, (1) schedule data input from the user terminal 30 (see FIG. 3), (2) entry and exit of users corresponding to the schedule information Time data (see FIG. 6) indicating the time of day and time work (see FIG. 6), (3) attribute data of the user corresponding to the schedule data (see FIG. 6), and (4) position data of the user's work location (see Based on four types of information and data, estimated schedule information and dashboard information shown below are generated and output.

First, as shown in FIG. 2, the schedule information (schedule data) inputted and generated by the schedule information input means 31 (scheduler) of the user terminal 30 is extracted and fetched by the schedule information management server 10 (step 01) The data is converted into a predetermined data format by the schedule information generation unit 11 and input to the administrator device 20 (schedule information reception unit 21).
For example, a predetermined number of pieces of schedule information such as “Department A: about 481 for about 560,000,” “Department B: for about 201, about 160,000,” are input to the administrator device 20.
An example of the schedule information extracted and input in this manner is shown in FIG.

  As shown in the figure, the schedule information input to the scheduler via the user terminal 30 is generated and output as schedule data in one row of XLS or CSV format for each “one event” (scheduled) (schedule Information generation means 11) is input to the administrator device 20 (schedule information reception means 21). In each row of schedule data, "user", "subject", "place", "start time", "end time", "host", "required attendee", "optional attendee", "reply" etc. Each item is included, and arbitrary text information is input.

Next, processing of machine learning by artificial intelligence is executed on the text information extracted in this manner (estimated schedule information generating means 22).
First, machine learning is performed on the text information extracted from the schedule information serving as the teacher data, with the predetermined number (for example, 10000) of the schedule information input to the administrator device 20 as the teacher data (Step 02 ).
Machine learning of teacher data, for example, makes 5,000 teacher data learn about 100 times.
After that, it is possible to verify the learning result for the remaining data (5000 cases).

Specifically, of text information extracted from schedule information to be teacher data, data indicating a sentence indicating “subject” is generated by morphological analysis.
Then, classification is performed by performing machine learning of a classification model based on a predetermined attribute on the decomposed data generated as such, and the schedule information is classified into a predetermined group unit, and the result is output as labeling information. It has become.
Here, “morphological analysis” is a natural language processing technology by a computer that divides a sentence written in a natural language into morphemes (minimum units having meaning in the language). A known technique can be used for such morphological analysis, and in the present embodiment, the administrator device 20 can be realized by being configured by a server computer provided with the function.

Also, with “machine learning”, useful rules, rules, judgment criteria, etc. are extracted from input data and repetitive learning is repeated to autonomously execute rules, rules, etc. for newly input data. It is an artificial intelligence technology that generates and grows on. The "class classification model" is a data mining technology that assigns target data to a target class.
Also for these machine learning and class classification models, known techniques can be used, and the administrator device 20 according to the present embodiment can be realized by being configured by a server computer or the like provided with those functions. .

In the present embodiment, as shown in FIG. 4, classification and analysis of text information decomposed as described above are performed by a method called, for example, RCNN (Recurrent Convolutional Neural Network).
RCNN is one of the models of neural networks.
In RCNN, RNN (Recurrent NN), which is also a model of neural network, can be put into the classifier up to the context, while the later words have a stronger effect on the learning result. It is reported that learning from two directions can eliminate bias and provide high classification accuracy.

So, in this embodiment, machine learning of the schedule information mentioned above is performed using RCNN.
Note that such RCNN can be implemented by an open source software library for performing calculation and learning of neural networks, such as, for example, Chainer.

FIG. 4 is an explanatory view showing a procedure of machine learning in RCNN.
In the figure, as an example of the subject included in the schedule information, the subject of “the final report meeting of company-wide R & D AI” is morpheme analysis / machine learning.
As shown in this figure, in RCNN, “morphometric analysis of the subject of the schedule” → “shake unique ID for each morpheme” → “allocation of numerical value to input node” → “activation” → “max pooling” The learning is performed according to the procedure of (maximum value extraction) → “softmax cross entropy” (abutment with the correct label, error calculation) → “error back propagation” (reverse calculation).
Such RCNN is improved in accuracy by combining the method of inputting in order from the back, instead of the simple recurrent type NW inputting in order from the front.

Then, using the teacher data for which a good verification result has been obtained by the RCNN as described above, the same RCNN applies to the remaining schedule information (for example, department A: 550,000, department B: 150,000) Machine learning is performed, and a predetermined labeling process is performed (step 03).
FIG. 5 shows an example of the labeling result by RCNN.
In the example shown in the figure, “0: in-house meeting”, “1: outside meeting”, “2: move”, “3: work”, “4: irrelevant”, “5: non-work” as the definition of labeling The five definitions of are set in advance, and based on these definitions, the corresponding subject is given a label corresponding to the subject included in the schedule information learned by the RCNN.
As a result, for the schedule included in each piece of schedule information, labeling information classified into group units according to the label definition as described above is generated.

Next, with respect to the schedule information (labeling information) labeled as described above, merge processing with the entry / exit data is performed (step 04).
FIG. 6 shows an example of the entrance / exit data.
Merging between entry and exit data as shown in the figure and labeling information is data indicating the time and place of entry (work) of each user to the office or the like and leaving (visiting), and the labeling information described above It is done by combining and integrating as one line of data.
For example, based on corresponding information such as “employee number”, “name”, “affiliate company”, “belonging department” and the like included in labeling information and entry / exit data, “date”, “day of week as entry / exit data Text information indicating “first entry time”, “last exit time”, “first entry gate”, “final exit gate” is integrated into labeling information and generated and output as one line of data It has become.

Next, processing is performed to convert the “place” information included in the labeling information merged with the entry / exit data as described above into latitude / longitude (step 05).
The conversion of the “place” information into position information (latitude / longitude information) is performed, for example, as “place” of the corresponding company as “position information of an office etc.” registered and held in advance in the manager device 20 described above. The latitude / longitude information such as the registered office, the in-house conference room, and the entrance / exit passage gate location is referred to and recorded in association with the corresponding labeling information.
FIG. 7 shows an example of “position information” registered in the present system 1.

As shown in FIG. 6A, for example, location name, address, latitude / longitude information of the head office or sales office of the corresponding company, a meeting room, a customer etc. is stored in advance in a storage means such as the manager device 20. Ru.
In addition, as shown in FIG. 6B, for example, for an arbitrary place name such as a place appearing frequently or a place name or an abbreviated name described as schedule information, identification information (“office number”) specifying the place Etc. can be registered in advance.
Such registration of position information is performed in advance manually.

Furthermore, as shown in FIG. 6C, for example, using a map search site (eg, "Google Map" etc.) or geocoding service (eg, GEO Coding API of "Google" etc.) published on the Internet Latitude / longitude information of an arbitrary “place” can be acquired and registered on the side of the administrator device 20.
Then, the “place” information included in the labeling information decomposed and extracted by the morphological analysis is converted into the latitude / longitude information registered / input to the administrator device 20 as described above, and the corresponding schedule information It is linked and recorded as position information.
This makes it possible to specify the "position" of the corresponding schedule.

Next, as described above, with regard to labeling information, positional information (latitude / longitude information) is added, and from the date / time / time, the positional relationship before and after the schedule, and the reply status included in the labeling information, etc. A process is performed to estimate the behavior of the subject (step 06).
This action estimation process takes in, for example, information in units of one day about estimated schedule information of each user to which merge of latitude / longitude information of machine learning / entry / exit data has been performed as described above, and It is a process of estimating the most certainty path from the event to the last event of the night.

Specifically, in the present embodiment, time information indicating the time when the user is going to work or leaving work, corresponding to the labeling information classified (labeled) in the predetermined group unit as described above by the estimated schedule information generation unit 22. And one of the schedule (labeling node) is extracted based on the predetermined rule based on the position information of the predetermined place including the work place of the user corresponding to the schedule information, and determined as the estimated schedule information .
FIG. 8 shows an example of logic for extracting a labeling node from overlapping schedule information and generating estimated schedule information.

In the example shown in the figure, the path with the highest certainty is extracted based on the following calculation rule, and is determined as estimated schedule information.
Calculation rule
・ The nodes of the first entry and the last exit always pass.
・ The node at the beginning of the day will be the first admission or the first meeting outside the company.
・ The node at the end of the day is the last exit, or the last meeting outside the company.
Continue searching for the most likely next node from the beginning of the day node to the end of the day node.
・ When moving time is considered, it does not go to unreachable nodes.
-If the schedule is duplicated (double booking, triple booking, etc.), refer to the reply status and determine the priority in the following order: "meeting organizer>consent> no reply> temporary schedule> no participation".
If priority is the same, prioritize nodes with earlier start times.

Based on the calculation rule as described above, in the example of FIG. 8, “first entry” is prioritized in the node at the beginning of the day, and “in-house meetings” overlapping at a time earlier than “first entry” are excluded. Be done.
In addition, in the node at the end of the day, "final leaving" and the last "outside meeting" take precedence, and "in-house meetings" overlapping at a later time than "final leaving" are excluded.
Further, in FIG. 8, with regard to a plurality of (two) “outside meetings” overlapping in the daytime daytime zone, the travel time is considered based on the position information associated with the corresponding schedule information. Exclude nodes that can not arrive at.
Also, with regard to such overlapping schedules, the highest priority schedule is adopted with reference to the reply status included in the corresponding schedule information, and the others are excluded.

According to the above logic, in the example shown in FIG. 8, the schedule following the nodes shown in bold is estimated to be the path with the highest certainty, and is determined as estimated schedule information for the current day of the user.
Then, the above-described behavior estimation process is repeated for each user's schedule information on a daily basis to generate and determine estimated schedule information for all the schedules.
The generated estimated schedule information is input to the dashboard display means 23, and a predetermined dashboard image is drawn (step 07). The dashboard image is displayed on a display or the like connected to any device / terminal such as the administrator device 20 or the user terminal 30, for example.

In the present embodiment, image information of an individual unit or a copy unit is displayed as a dashboard image, in which the estimated schedule information is listed in order of date and time in units of a predetermined period (see FIG. 9).
In addition, generation of a dashboard image by the dashboard display means 23 converts, for example, data in XLS or CSV format, which is schedule information, into data in XML format of a JSON file, for example, a known application such as “Bootstrap” The framework can generate the desired dashboard format image

Specifically, examples of dashboard images generated by the dashboard display means 23 include drawing data as shown in FIGS.
The dashboard image shown in FIG. 9 is estimated as the schedule information for each user, on the left side of the image, the original schedule information input to the scheduler by the user, and on the right side of the image, the system generated. Each schedule information is displayed in a calendar format. In such calendar information, the list display format can be changed in units of day, week, month, etc. according to an input operation or the like.

Then, as shown in FIG. 9, the estimated schedule information on the right side of the image is, for example, “in-house meeting”, “outside meeting”, “move”, “work, etc.”, “group” classified by the above-described labeling process. The first entry and the last exit can be displayed with backgrounds, borders, etc. of different colors and densities, and the contents of the schedule can be recognized and identified at a glance.
Further, in the estimated schedule information displayed as described above, one piece of information (see FIG. 8) “estimated” is displayed for a plurality of schedules whose time zones overlap with each other in the original schedule information.

Furthermore, in the estimated schedule information, a schedule of, for example, “move” or “work, etc.” is indicated as the information “estimated” for the blank time zone which was not included in the original schedule information.
Also, in this dashboard image, the upper row of the image shows “average working hours per day”, “average opening hours”, “average closing hours”, “working days over 10 hours”, “number of meetings outside the company”, Items such as "the number of meetings in the company" are displayed, and the corresponding data is displayed.
These data are extracted and generated by the estimated schedule information generation means 22 based on the information of the scheduler and the entrance / exit data input to the manager device 20 side.

As described above, in the dashboard image generated by the dashboard display unit 23, information indicating how the individual unit of each user works is visualized, and in addition, the entire group unit or company unit to which each user belongs The way of working is also visualized as image information.
Specifically, in the example shown in FIG. 10, the way of working of each user is visualized as a dashboard image, and "average day", "general state of working", "original schedule information" , "Artificial intelligence and classified schedule information" can now be grasped visually at a glance.
Also, in the example shown in FIG. 11, the way of working of the department unit to which the target user belongs and the whole organization is visualized. For example, the “overall overview” of the corresponding company, “analysis of way of working” (activity time classification) , Information such as "Individual Ranking" (long-time observers), "Department Ranking" (working hours by department), "Analysis of personal choice", "Distribution of employees in internal and external meetings" can be recognized at a glance It is supposed to be.

The dashboard images shown in FIGS. 9 to 11 are an example of dashboard information that can be generated and output by the present system 1, and it goes without saying that other data can be generated and output as dashboard images.
That is, the various data input to the above-described administrator device 20, (1) schedule data input to the scheduler, (2) time data of user entry / exit (going to / from work) (see FIG. 6), (3) If it is information that can be analyzed and tabulated based on four types of information such as user attribute data (see FIG. 6) and (4) position data of the user's work location (see FIG. 7), the function of the user or organization It is possible to generate it as dashboard information which visualized one.

As described above, according to the present system 1, predetermined estimated schedule information is generated by taking in schedule information input to the scheduler via the user terminal 30 or the like and sorting by machine learning by artificial intelligence. Since it is possible to use the artificial intelligence to sort the various information input to the scheduler using artificial intelligence, instead of simply making the scheduler easy to view and to display related information, how to use working hours, distribution, trends, etc. It can be reconstructed as analyzed estimated schedule information.
Then, the schedule information analyzed in such a manner can be generated as visualized dashboard information such that the contents can be recognized and grasped at a glance.

For example, according to the present system, “start time,” “work time,” “working time,” “number of internal meeting times / hours,” “number of outside meeting times, time,” “moving time,” “time of work, etc.” Elements such as “Ranking” and “Activity location” can be sorted by artificial intelligence and analyzed, and the analysis results can be visualized as dashboard information, so by using that information as an analysis tool, how to work You can hit the point of reform.
Specifically, by analyzing the raw data shown as dashboard information, it is possible to construct the following hypotheses for working style reform.
・ Classification of meeting contents: Project unit time ・ Many unit time of work unit ・ Movement time classification: Analysis of appearance place ・ Work with high seasonality: Analysis by month ・ season by year ・ Work different by job title / class: Analysis by class ・ Analysis by class Inward / outward assessment: A meeting analysis with people inside and outside the department

And, by conducting analysis in the additional survey, it is possible to test the above hypothesis.
・ Detailed grasp of the meeting ・ Improved proposal: Detailed analysis of the work content by interview ・ Questionnaire ・ Detailed grasp of the work ・ Improvement plan: Detailed analysis of the work content by interview ・ Questionnaire ・ Competency model making: Correspondence with the personnel appraisal section Furthermore, From the analysis like the above, it is possible to narrow down on the important issues and clarify the hitter according to the issues.
・ Work efficiency improvement ・ Reduction of meeting time ・ Reduction of the number of meetings ・ Reduction of travel time

As described above, according to the present system 1, it is possible to realize grasping and analysis of the working hours and working conditions, reform of the working style of individuals and the whole work, improvement of productivity and improvement of the working environment, etc. It will be.
Although the preferred embodiments of the present invention have been shown and described above, the present invention is not limited to the above-described embodiments, and it goes without saying that various modifications can be made within the scope of the present invention.
For example, in the embodiment described above, “in-house meeting”, “outside meeting”, “move”, “work”, “irrelevant”, “non-work”, etc. are defined as the labeling to classify each schedule information into group units. Although this is shown as an example of grouping, any classification / definition can be set / adopted according to, for example, the type of work of the corresponding company and the contents of work of the type and label definition of the group to be classified.

Moreover, in the embodiment described above, an example of the user individual and the whole department / organization is shown as working method information that can be visualized by a dashboard image (see FIGS. 9 to 11). It can be visualized by
For example, it is possible to extract "interest" in individual units from the subject and content included in schedule information and visualize as image information that arranges the arrangement, font, size, color, etc. of characters according to the frequency of occurrence. It is.
In addition, from the subject and content included in the schedule information, information on who is meeting with whom and how often, and how often they are meeting is extracted, and the image of connecting person names in a three-dimensional space form Can also be visualized.

  The present invention can be suitably used for a service for recording and managing work schedules of workers and employees using schedule software, an in-house system, and the like.

DESCRIPTION OF SYMBOLS 1 Schedule information analysis system 10 Schedule information management server 11 Schedule information generation means 20 Administrator apparatus 21 Schedule information reception means 22 Estimated schedule information generation means 23 Dashboard display means 30 User terminal 31 Schedule information input means

Claims (4)

A schedule information analysis system that generates predetermined estimated schedule information based on schedule information indicating a user's scheduled content and scheduled date and time.
Schedule information receiving means for receiving the input of the schedule information;
Estimate schedule information generation means for classifying the schedule information into predetermined group units and generating predetermined estimated schedule information by performing machine learning on a subject sentence included in the schedule information. Schedule information analysis system. The estimated schedule information generation means
Time information indicating the time of day attendance and leaving of the user corresponding to the schedule information;
Based on the position information of the user corresponding to the schedule information,
When the schedule information classified into a predetermined group unit includes a plurality of pieces of schedule information overlapping in time zones, any one piece of schedule information is extracted to generate the estimated schedule information. The schedule information analysis system according to claim 1. A dashboard display unit configured to generate, as predetermined dashboard information, image information of an individual unit or a copy unit in which the estimated schedule information is listed and displayed in order of date and time with a predetermined period as a unit;
The schedule information analysis system according to claim 1 or 2, characterized in that: An information display device constituting a schedule information analysis system that generates predetermined estimated schedule information based on schedule information indicating a user's scheduled content and scheduled date and time;
Schedule information receiving means for receiving the input of the schedule information;
It is characterized in that by subjecting subject sentences contained in the schedule information to machine learning, the schedule information is classified into a predetermined group unit to function as estimated schedule information generating means for generating predetermined estimated schedule information. Schedule information analysis program.