Operation log acquisition device and method

Abstract

An operation log acquisition device (1) according to one embodiment comprises: a detection unit (311) that detects an operation event when a user performs an operation on a screen including a plurality of operation components and a plurality of pieces of label information, and identifies a first operation component operated among the operation components; a calculation unit (316) that calculates the degree of similarity between an ID of the first operation component and each of the pieces of label information; an estimation unit (317) that estimates that first label information having the highest degree of similarity to the ID among the calculated degrees of similarity corresponds to the ID; and a creation unit (318) that creates operation log information INF_G including the ID and the first label information.

Description

Operation log acquisition device and method

[0001] One aspect of the present invention relates to an operation log acquisition device and method.

[0002] In recent years, many companies have begun promoting digital transformation (DX), which involves using data and digital technology to transform their operations. Generally, when promoting business transformation, the first step is to understand the current state of operations.

[0003] Non-patent document 1 describes a log acquisition tool as a technology for assisting in understanding the current state of business operations. The tool captures the timing at which a terminal user operates information input fields, buttons, etc. when operating a business system, etc., and records the operations performed by the terminal user as an operation log.

[0004] Taisuke Wakasugi et al., "Operation Process Classification-Based Business Design Support Technology Contributing to the Promotion of DX," NTT Technical Journal, Internet <URL: https: / / journal.ntt.co.jp / article / 18032>, [Retrieved November 10, 2023]

[0005] The information recorded as an operation log by the log acquisition tool includes the date and time (operation time), business screen information (window title), and an ID (operation component ID) that identifies the operated operation component (text box, button, etc.). Therefore, by analyzing the operation log, it is possible to understand what operation the terminal user performed on which screen of the business system, in what procedure and timing. This allows for various business analyses of terminal tasks.

[0006] However, when analyzing a business using operation logs, analysts need to refer to the screen data of the business system from which the logs were acquired and compare the operation part IDs with the screen data of the business system in order to understand the operations of the terminal user. Therefore, if the system from which the logs were acquired is updated and the ID system changes, this comparison (matching) becomes difficult.

[0007] This invention has been made with the above-mentioned circumstances in mind, and provides an operation log acquisition device and method that makes it possible to understand the operations of a terminal user even if the system from which the log is acquired has been updated and the ID system is different.

[0008] In order to solve the above problem, an operation log acquisition device according to one embodiment of the present invention includes a detection unit that detects an operation event when a user performs an operation on a screen that includes a plurality of operation components and a plurality of label information, and identifies a first operation component that was operated among the operation components; a calculation unit that calculates a similarity between the ID of the first operation component and each piece of label information; an estimation unit that estimates that the first label information that has the highest similarity to the ID among the calculated similarities corresponds to the ID; and a creation unit that creates operation log information that includes the ID and the first label information.

[0009] An operation log acquisition method according to one embodiment of the present invention includes detecting an operation event when a user performs an operation on a screen including a plurality of operation components and a plurality of label information, identifying a first operation component that was operated among the operation components, calculating a similarity between the ID of the first operation component and each piece of label information, inferring that a piece of first label information that has the highest similarity to the ID among the calculated similarities corresponds to the ID, and creating operation log information that includes the ID and the first label information.

[0010] According to one aspect of the present invention, even if the system from which the log is acquired is updated and the ID system is changed, the operation of the terminal user can be grasped.

[0011] FIG. 1 is a block diagram showing an example of the configuration of an operation log acquisition device according to an embodiment. FIG. 2 is a block diagram showing an example of the hardware configuration of an operation log acquisition device according to an embodiment. FIG. 3 is a block diagram showing an example of the functional configuration of an operation log acquisition device according to an embodiment. FIG. 4 is a diagram showing an example of a business system screen on which the operation log acquisition device according to an embodiment performs operation log acquisition processing. FIG. 5 is a diagram showing an example of the relationship between operation object IDs and word vectors of the operation object IDs digitized by a word digitization process of the operation log acquisition device according to an embodiment. FIG. 6 is a diagram showing an example of the relationship between label information and word vectors of the label information digitized by a word digitization process of the operation log acquisition device according to an embodiment. FIG. 7 is a diagram explaining an example of a similarity calculation process of the operation log acquisition device according to an embodiment. FIG. 8 is a flowchart showing an example of an operation log acquisition process of the operation log acquisition device according to an embodiment.

[0012] Hereinafter, embodiments will be described with reference to the drawings.

[0013] 1. Embodiment An operation log acquisition device according to an embodiment will be described. The following describes an example of an operation log acquisition device that acquires operation logs when a user operates a screen (window) of a business system displayed in a web browser. Hereinafter, the screen of the business system will be referred to as a "business system screen."

[0014] 1.1 Configuration 1.1.1 Configuration of Operation Log Acquisition Device The configuration of the operation log acquisition device according to this embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the configuration of the operation log acquisition device.

[0015] The operation log acquisition device 1 is a device that identifies an operation component in a business system screen operated by a user, estimates label information (label name) in the business system screen that corresponds to the operation component ID of the identified operation component, and acquires operation log information including the operation component ID and the estimated label information. The operation component ID is a code for performing information processing and identifies an information input item or button (operation component) in the business system screen operated by the user. The label information is information for explaining the information input item or button to the user.

[0016] As shown in FIG. 1, the operation log acquisition device 1 includes an input unit 2, a processing unit 3, and a display unit 4.

[0017] The input unit 2 is a user interface such as a mouse, a keyboard, a touch panel, etc. The input unit 2 provides the processing unit 3 with instructions from the user to the business system (such as input of information, selection from a list, and clicking of a button).

[0018] The processing unit 3 is, for example, a computer. The processing unit 3 starts a web browser based on a user instruction received from the input unit 2. The processing unit 3 then causes the display unit 4 to display a business system screen via the web browser.

[0019] The processing unit 3 also identifies an operation component on the business system screen that has been operated by the user. The processing unit 3 estimates label information on the business system screen that corresponds to the operation component ID of the identified operation component. The processing unit 3 acquires operation log information that includes the operation component ID and the estimated label information.

[0020] The display unit 4 is, for example, a monitor, and displays a business system screen via a web browser.

[0021] 1.1.2 Hardware Configuration of Operation Log Acquisition Device The hardware configuration of the operation log acquisition device 1 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing an example of the hardware configuration of the operation log acquisition device 1. Fig. 2 also shows an external operation log storage device 5 that is communicably connected to the operation log acquisition device 1 via a network.

[0022] As shown in FIG. 2 , the processing unit 3 includes a processor 31 , a memory 32 , and a communication interface 33 .

[0023] The processor 31 is, for example, a central processing unit (CPU), a micro processing unit (MPU), a graphics processing unit (GPU), or a field programmable gate array (FPGA). The processor 31 executes, for example, a process for acquiring operation log information. Hereinafter, the process for acquiring operation log information will be referred to as an "operation log acquisition process."

[0024] The operation log acquisition process includes an event detection process, a screen data acquisition process, an operation information acquisition process, a label information acquisition process, a word quantification process, a similarity calculation process, a label information estimation process, and a log information creation process. Details of the event detection process, the screen data acquisition process, the operation information acquisition process, the label information acquisition process, the word quantification process, the similarity calculation process, the label information estimation process, and the log information creation process will be described later.

[0025] The memory 32 includes, for example, a read-only memory (ROM) and a random access memory (RAM). The ROM stores, for example, a program for causing the processor 31 to execute the operation log acquisition process. The RAM is used as a working area for the processor 31. The RAM temporarily stores, for example, the above-mentioned program executed by the processor 31 and data during the execution of the operation log acquisition process.

[0026] The communication interface 33 controls communication between the input unit 2, the display unit 4, and the operation log storage device 5 and the processing unit 3. The communication interface 33 receives instructions from the user from the input unit 2. The communication interface 33 transmits an image of the business system screen to the display unit 4. The communication interface 33 transmits operation log information to the operation log storage device 5.

[0027] The operation log storage device 5 is, for example, a storage medium used to store data, programs, etc. The operation log storage device 5 stores, for example, operation log information. Note that the operation log storage device 5 may be included in the operation log acquisition device 1.

[0028] 1.1.3 Functional Configuration of Operation Log Acquisition Device The functional configuration of the operation log acquisition device 1 will be described with reference to Fig. 3. Fig. 3 is a block diagram showing an example of the functional configuration of the operation log acquisition device 1. Fig. 3 also shows users and the operation log storage device 5.

[0029] 3, the processing unit 3 includes, as functional blocks, an event detection unit 311, a screen data acquisition unit 312, an operation information acquisition unit 313, a label information acquisition unit 314, a word digitization unit 315, a similarity calculation unit 316, a label information estimation unit 317, and a log information creation unit 318. The processor 31 of the processing unit 3 functions as the event detection unit 311, the screen data acquisition unit 312, the operation information acquisition unit 313, the label information acquisition unit 314, the word digitization unit 315, the similarity calculation unit 316, the label information estimation unit 317, and the log information creation unit 318. Note that, in FIG. 3, functional blocks corresponding to the memory 32 and the communication interface 33 of the processing unit 3 are omitted from the illustration.

[0030] The following describes an example in which a user uses a keyboard to enter information into a name input text box on a business system screen that includes operational components such as a text box for entering a name and a text box for entering a telephone number (telephone number). Hereinafter, the text box for entering a name will be referred to as the "name text box."

[0031] (Event Detection Unit 311) The event detection unit 311 executes an event detection process, which is a process of detecting an operation event when a user performs an operation such as inputting information, selecting from a list, or clicking a button on a business system screen that includes a plurality of operation components and a plurality of pieces of label information, and identifying the operated operation component.

[0032] For example, the event detection unit 311 acquires (captures) images of the business system screen at predetermined intervals. When keyboard input is performed in a text box on the business system screen, the event detection unit 311 detects the occurrence of a keyboard input event by using a function provided by the OS (Operating System). The event detection unit 311 then compares the images of the business system screen before and after the event occurred and detects differences between these images to identify the operational component on the business system screen that was operated by the user.

[0033] When a user inputs information into the name text box using the keyboard, the event detection unit 311 detects the occurrence of a keyboard input event. The event detection unit 311 then compares an image of the business system screen before the event occurred, in which nothing was entered in the name text box, with an image of the business system screen after the event occurred, in which information was entered in the name text box, and detects any differences between these images. Next, the event detection unit 311 identifies the operational component located at that position based on the position on the business system screen where the difference exists. As a result, the event detection unit 311 identifies the name text box as the operational component operated by the user.

[0034] For example, when a drop-down list is selected or a button is pressed within a business system screen, the event detection unit 311 uses a function provided by the OS to detect the occurrence of the event of selecting a drop-down list or pressing a button, and identifies the operation component operated by the user.

[0035] When the event detection unit 311 detects the occurrence of an event (identifies an operation component), it transmits a notification indicating that the event has been detected to the screen data acquisition unit 312 and the operation information acquisition unit 313. Hereinafter, the notification indicating that an event has been detected will be referred to as a "detection notification DN." Note that the event detection unit 311 may transmit information on the identified operation component (for example, an operation component ID) to the screen data acquisition unit 312 and the operation information acquisition unit 313 instead of the detection notification DN.

[0036] (Screen Data Acquisition Unit 312) The screen data acquisition unit 312 executes a screen data acquisition process. The screen data acquisition process is a process for acquiring data of a business system screen operated by a user. Hereinafter, the data of a business system screen operated by a user will be referred to as "screen data DT."

[0037] For example, when the screen data acquisition unit 312 receives a detection notification DN from the event detection unit 311, it acquires screen data DT of the business system screen operated by the user. The screen data DT is, for example, HTML data (index.html) of the business system screen. The screen data acquisition unit 312 transmits the acquired screen data DT to the operation information acquisition unit 313 and the label information acquisition unit 314.

[0038] (Operation Information Acquisition Unit 313) The operation information acquisition unit 313 executes operation information acquisition processing. The operation information acquisition processing is processing for acquiring operation information related to an operation event detected by the event detection processing. Hereinafter, operation information related to an operation event will be referred to as "operation information INF_O." The operation information INF_O includes, for example, an operation component ID. Note that the operation information INF_O may also include an operation time, a window title, and a URL.

[0039] For example, when the operation information acquisition unit 313 receives a detection notification DN from the event detection unit 311, it receives screen data DT from the screen data acquisition unit 312. Then, the operation information acquisition unit 313 acquires an operation object ID as operation information INF_O based on the received screen data DT.

[0040] When the user inputs information into the name text box using the keyboard, the operation information acquisition unit 313 acquires the HTML tag information (<input id="name" type="text" / > Then, the operation information acquisition unit 313 acquires the operation component ID ("name") based on the acquired HTML tag information.

[0041] The operation information acquisition unit 313 transmits the acquired operation object ID to the word digitization unit 315 and the log information creation unit 318 .

[0042] (Label Information Acquisition Unit 314) The label information acquisition unit 314 executes label information acquisition processing. The label information acquisition processing is processing for acquiring each piece of label information in the business system screen based on the screen data DT acquired by the screen data acquisition processing. Hereinafter, the acquired label information will be referred to as "label information INF_L." A list of all label information INF_L will be referred to as "label information list LST_L." The label information list LST_L includes one or more pieces of label information INF_L.

[0043] For example, the label information acquisition unit 314 receives screen data DT from the screen data acquisition unit 312. Then, based on the received screen data DT, the label information acquisition unit 314 acquires a list of all label information INF_L in the business system screen (label information list LST_L).

[0044] In the example of FIG. 3, the label information acquisition unit 314 acquires a label information list LST_L including "Name", "Telephone", . . . as label information INF_L based on the screen data DT.

[0045] The label information acquisition unit 314 transmits the acquired label information list LST_L to the word digitization unit 315 .

[0046] (Word Quantification Unit 315) The word quantification unit 315 executes a word quantification process. The word quantification process is a process of quantifying (word vectorizing) the operation object ID (character string) acquired by the operation information acquisition process and each piece of label information INF_L (character string) in the label information list LST_L acquired by the label information acquisition process, using a distributed representation of a word vector. Hereinafter, the quantified word vector of the operation object ID will be referred to as a "word vector VCT_P." The quantified word vector of each piece of label information INF_L will be referred to as a "word vector VCT_L." A list of word vectors VCT_L corresponding to all pieces of label information INF_L will be referred to as a "word vector list LST_W." The word vector list LST_W includes one or more word vectors VCT_L.

[0047] For example, the word quantification unit 315 receives an operation object ID from the operation information acquisition unit 313. Then, the word quantification unit 315 divides the received operation object ID into one or more words. After that, the word quantification unit 315 quantifies each of the divided one or more words using distributed representation of word vectors, and combines the quantified word vectors to quantify the operation object ID as a word vector.

[0048] Furthermore, the word quantification unit 315 receives the label information list LST_L from the label information acquisition unit 314. Then, the word quantification unit 315 divides each piece of label information INF_L in the received label information list LST_L into one or more words. Thereafter, the word quantification unit 315 quantifies each of the divided one or more words using distributed representations of word vectors, and combines the quantified word vectors to quantify each piece of label information INF_L as a word vector.

[0049] In the example of FIG. 3 , the word quantification unit 315 divides the operation object ID ("name") into one or more words. As a result, one word "name" is obtained. Then, the word quantification unit 315 quantifies "name" using a distributed representation of the word vector, thereby obtaining, for example, {1, 2, 3, 4} as the word vector VCT_P of "name."

[0050] Furthermore, the word quantification unit 315 divides each of all label information INF_L ("Name", "Telephone", ...) in the label information list LST_L into one or more words. As a result, one word "Name" is obtained for the label information INF_L ("Name"). One word "Telephone" is obtained for the label information INF_L ("Telephone"). Similarly, one or more words are obtained for the other label information INF_L. Then, for the label information INF_L ("Name"), the word quantification unit 315 quantifies "Name" using the distributed representation of the word vector, thereby obtaining, for example, {1, 2, 3, 4} as the word vector VCT_L of "Name". For the label information INF_L ("Telephone"), the word quantification unit 315 quantifies "Telephone" using the distributed representation of the word vector, thereby obtaining, for example, {5, 6, 7, 8} as the word vector VCT_L of "Telephone". Similarly, word vectors VCT_L are obtained for other label information INF_L. As a result, a word vector list LST_W is obtained that includes {1, 2, 3, 4}, {5, 6, 7, 8}, . . . as word vectors VCT_L.

[0051] The word digitizing unit 315 transmits the digitized word vector VCT_P of the operation object ID and the word vector list LST_W corresponding to the label information list LST_L to the similarity calculation unit 316 .

[0052] (Similarity Calculation Unit 316) The similarity calculation unit 316 executes a similarity calculation process. The similarity calculation process calculates the similarity between the operation object ID and each piece of label information INF_L in the label information list LST_L. More specifically, the similarity calculation process calculates the similarity between the word vector VCT_P of the operation object ID obtained by the word quantification process and each piece of word vector VCT_L in the word vector list LST_W using the inter-vector angle, i.e., cosine similarity. Hereinafter, each calculated similarity will be referred to as a "similarity DEG_S." Information including the similarity DEG_S and the operation object ID and label information INF_L corresponding to the similarity DEG_S will be referred to as "similarity information INF_S." A list of all similarity information INF_S will be referred to as a "similarity list LST_S." The similarity list LST_S includes one or more pieces of similarity information INF_S.

[0053] For example, the similarity calculation unit 316 receives the word vector VCT_P and the word vector list LST_W from the word quantification unit 315. Then, the similarity calculation unit 316 calculates the similarity DEG_S between the word vector VCT_P of the operation object ID and each of the word vectors VCT_L in the word vector list LST_W by using cosine similarity.

[0054] That is, the calculation by the similarity calculation unit 316 includes calculating the similarity DEG_S between the word vector VCT_P of the digitized operation object ID and each word vector VCT_L of the digitized label information INF_L using cosine similarity.

[0055] In the example of FIG. 3 , the similarity calculation unit 316 calculates the similarity DEG_S between the word vector VCT_P ({1, 2, 3, 4}) of "name" and the word vector VCT_L ({1, 2, 3, 4}) of "full name" by using cosine similarity. The similarity DEG_S is, for example, a value obtained by converting the cosine similarity normalized to a range of -1 to 1 into a range of 0 to 100. As a result, the similarity DEG_S between "name" and "full name" is obtained as, for example, 100. Furthermore, the similarity information INF_S between "name" and "full name" is obtained as "name × full name = 100." The similarity calculation unit 316 calculates the similarity DEG_S between the word vector VCT_P of "name" and the word vector VCT_L ({5, 6, 7, 8}) of "telephone" by using cosine similarity. As a result, for example, 30 is obtained as the similarity DEG_S between "name" and "telephone". Furthermore, "name × telephone = 30" is obtained as the similarity information INF_S between "name" and "telephone". Similarly, similarity DEG_S and similarity information INF_S are obtained for the other word vectors VCT_L. As a result, a similarity list LST_S is obtained that includes "name × full name = 100", "name × telephone = 30", ... as the similarity information INF_S.

[0056] The similarity calculation unit 316 transmits the calculated similarity list LST_S to the label information estimation unit 317 .

[0057] (Label Information Estimation Unit 317) The label information estimation unit 317 executes label information estimation processing. The label information estimation processing is processing for estimating label information INF_L corresponding to an operation object ID based on the similarity list LST_S obtained by the similarity calculation processing.

[0058] For example, the label information estimation unit 317 receives the similarity list LST_S from the similarity calculation unit 316. Then, the label information estimation unit 317 estimates label information INF_L corresponding to the operation object ID based on the received similarity list LST_S. More specifically, the label information estimation unit 317 estimates that the label information INF_L having the highest similarity DEG_S with the operation object ID, among the calculated similarities DEG_S included in the similarity information INF_S in the similarity list LST_S, corresponds to the operation object ID. Hereinafter, the estimated label information INF_L will be referred to as "label information INF_Le."

[0059] In the example of FIG. 3 , the label information estimation unit 317 refers to each piece of similarity information INF_S in the similarity list LST_S. Of the calculated similarities DEG_S included in the similarity information INF_S in the similarity list LST_S, 100 is the highest. Therefore, the label information estimation unit 317 selects the similarity information INF_S ("name × full name = 100"). That is, the label information estimation unit 317 estimates that the label information INF_L ("full name") is the label information INF_Le corresponding to the operational object ID ("name").

[0060] The label information estimation unit 317 transmits the estimated label information INF_Le to the log information creation unit 318 .

[0061] (Log information creation unit 318) The log information creation unit 318 executes a log information creation process. The log information creation process is a process for creating operation log information including the operation object ID acquired by the operation information acquisition process and the label information INF_Le corresponding to the operation object ID estimated by the label information estimation process. Hereinafter, the created operation log information will be referred to as "operation log information INF_G."

[0062] For example, the log information creation unit 318 receives an operation object ID from the operation information acquisition unit 313. The log information creation unit 318 receives label information INF_Le from the label information estimation unit 317. Then, the log information creation unit 318 creates operation log information INF_G including the received operation object ID and label information INF_Le. As a result, the operation log acquisition device 1 acquires the operation log information INF_G.

[0063] In the example of FIG. 3, the log information creation unit 318 creates operation log information INF_G including the operation object ID ("name") and label information INF_Le ("name").

[0064] The log information creating unit 318 transmits the created operation log information INF_G to the operation log storage device 5 .

[0065] Next, a description will be given of the word quantification process, similarity calculation process, and label information estimation process when a user performs an operation on the business system screen shown in FIG.

[0066] FIG. 4 is a diagram showing an example of a business system screen on which the operation log acquisition device 1 performs the operation log acquisition process.

[0067] The business system screen shown in FIG. 4 includes a member code text box, a member name text box, a contact information text box, and a search button as operation components related to searches. Label information INF_L describing the operation components includes "member code," "member name," "contact information," and "search." The operation component ID of the member code text box is, for example, "member_code." The operation component ID of the member name text box is, for example, "member_name." The operation component ID of the contact information text box is, for example, "contact1." The operation component ID of the search button is, for example, "search."

[0068] The business system screen shown in FIG. 4 also includes operation components related to registering detailed information, such as an account text box, an address text box, a contact text box, a transaction amount text box, and a register button. Label information INF_L describing the operation components includes "account," "address," "contact," "transaction amount," and "register." The operation component ID of the account text box is, for example, "account." The operation component ID of the address text box is, for example, "address." The operation component ID of the contact text box is, for example, "contact2." The operation component ID of the transaction amount text box is, for example, "trade_price." The operation component ID of the register button is, for example, "submit."

[0069] First, the word digitization process will be described.

[0070] In the example of FIG. 4 , the word quantification unit 315 divides the operation object ID (“member_code”) into one or more words. As a result, two words “member” and “code” are obtained. Thereafter, the word quantification unit 315 quantifies “member” and “code” using a distributed representation of the word vector, and combines the quantified word vectors of “member” and “code” to obtain, for example, {6, 1, 3, 3} as the word vector VCT_P of “member_code” as shown in FIG. 5 . FIG. 5 is a diagram showing an example of the relationship between the operation object ID and the word vector VCT_P of the operation object ID quantified by the word quantification process.

[0071] Similarly, word vectors VCT_P are obtained for other operation part IDs. As shown in FIG. 5, for example, {6, 1, 3, 7} is obtained as the word vector VCT_P for the operation part ID ("member_name"). For example, {4, 1, 3, 4} is obtained as the word vector VCT_P for the operation part ID ("contact1"). For example, {1, 4, 1, 1} is obtained as the word vector VCT_P for the operation part ID ("search"). Note that when the number of words obtained by division is one, such as "search," the word vector obtained by combining is the word vector of "search."

[0072] As shown in FIG. 5 , for example, {3, 3, 3, 1} is obtained as the word vector VCT_P of the operation part ID ("account"). For example, {4, 1, 5, 2} is obtained as the word vector VCT_P of the operation part ID ("address"). For example, {4, 1, 2, 4} is obtained as the word vector VCT_P of the operation part ID ("contact2"). For example, {2, 9, 2, 6} is obtained as the word vector VCT_P of the operation part ID ("submit"). For example, {1, 5, 1, 1} is obtained as the word vector VCT_P of the operation part ID ("submit").

[0073] In the example of FIG. 4 , the word digitization unit 315 divides the label information INF_L ("membership code") into one or more words. This results in two words, "member" and "code." The word digitization unit 315 then digitizes "member" and "code" using a distributed representation of the word vector, and combines the digitized word vectors of "member" and "code" to obtain, for example, {6, 2, 3, 3} as the word vector VCT_L of "membership code," as shown in FIG. 6 . FIG. 6 is a diagram showing an example of the relationship between the label information INF_L and the word vector VCT_L of the label information INF_L digitized by the word digitization process.

[0074] Similarly, word vectors VCT_L are obtained for other label information INF_L. As shown in FIG. 6, for example, {6, 2, 3, 7} is obtained as the word vector VCT_L of label information INF_L ("member name"). For example, {4, 2, 3, 4} is obtained as the word vector VCT_L of label information INF_L ("contact information"). For example, {1, 5, 1, 1} is obtained as the word vector VCT_L of label information INF_L ("search"). Note that when the number of words obtained by division is one, such as "search", the word vector obtained by combining is the word vector of "search".

[0075] 6, for example, {3, 4, 3, 1} is obtained as the word vector VCT_L of the label information INF_L ("account"). For example, {4, 2, 5, 2} is obtained as the word vector VCT_L of the label information INF_L ("address"). For example, {4, 2, 2, 4} is obtained as the word vector VCT_L of the label information INF_L ("contact information"). For example, {2, 0, 2, 6} is obtained as the word vector VCT_L of the label information INF_L ("transaction amount"). For example, {1, 6, 1, 1} is obtained as the word vector VCT_L of the label information INF_L ("registration").

[0076] Next, the similarity calculation process will be described.

[0077] In the example of FIG. 4 , consider a case where the event detection unit 311 identifies the member code text box as the operation component. In this case, the operation information acquisition unit 313 acquires "member_code" as the operation component ID. As shown in FIG. 7 , the similarity calculation unit 316 calculates the similarity DEG_S between the word vector VCT_P ({6, 1, 3, 3}) of the operation component ID ("member_code") and the word vector VCT_L ({6, 2, 3, 3}) of the label information INF_L ("member code"). FIG. 7 is a diagram illustrating an example of the similarity calculation process.

[0078] Similarly, as shown in FIG. 7 , the similarity DEG_S between the word vector VCT_P of the operation object ID (“member_code”) and the word vector VCT_L ({6, 2, 3, 7}) of the label information INF_L (“member name”) is calculated. The similarity DEG_S between the word vector VCT_P of the operation object ID (“member_code”) and the word vector VCT_L ({4, 2, 3, 4}) of the label information INF_L (“contact information”) is calculated. The similarity DEG_S between the word vector VCT_P of the operation object ID (“member_code”) and the word vector VCT_L ({1, 5, 1, 1}) of the label information INF_L (“search”) is calculated.

[0079] 7, a similarity DEG_S between the word vector VCT_P of the operation object ID ("member_code") and the word vector VCT_L ({3, 4, 3, 1}) of the label information INF_L ("account") is calculated. A similarity DEG_S between the word vector VCT_P of the operation object ID ("member_code") and the word vector VCT_L ({4, 2, 5, 2}) of the label information INF_L ("address") is calculated. A similarity DEG_S between the word vector VCT_P of the operation object ID ("member_code") and the word vector VCT_L ({4, 2, 2, 4}) of the label information INF_L ("contacts") is calculated. The similarity DEG_S between the word vector VCT_P of the operation item ID ("member_code") and the word vector VCT_L ({2, 0, 2, 6}) of the label information INF_L ("transaction amount") is calculated. The similarity DEG_S between the word vector VCT_P of the operation item ID ("member_code") and the word vector VCT_L ({1, 6, 1, 1}) of the label information INF_L ("registration") is calculated.

[0080] When the operation part ID is other than "member_code", the similarity DEG_S is calculated in the same manner.

[0081] The label information estimation process is similar to the example in FIG.

[0082] 1.2 Operation Log Acquisition Processing The operation log acquisition processing will be described with reference to Fig. 8. Fig. 8 is a flowchart showing an example of the operation log acquisition processing of the operation log acquisition device 1.

[0083] When the operation log acquisition process is started, the event detection unit 311 starts acquiring images of the business system screen at predetermined intervals.

[0084] When the event detection unit 311 detects the occurrence of an event (identifies an operation component) as described above (S101_Yes), the event detection unit 311 transmits a detection notification DN to the screen data acquisition unit 312 and the operation information acquisition unit 313.

[0085] On the other hand, if the event detection unit 311 does not detect the occurrence of an event (S101_No), the process returns to step S101.

[0086] Next, when the screen data acquisition unit 312 receives the detection notification DN from the event detection unit 311, it acquires screen data DT of the business system screen operated by the user (S102).

[0087] Next, when the operation information acquisition unit 313 receives the detection notification DN from the event detection unit 311, it receives the screen data DT from the screen data acquisition unit 312 and acquires the operation item ID based on the screen data DT as described above (S103).

[0088] Next, the label information acquisition unit 314 receives the screen data DT from the screen data acquisition unit 312, and acquires each piece of label information INF_L in the business system screen based on the screen data DT as described above (S104).

[0089] Next, the processing unit 3 calculates the similarity DEG_S between the operation object ID and each piece of label information INF_L. More specifically, the word quantification unit 315 quantifies each of the operation object ID and the label information INF_L using the distributed representation of the word vectors as described above (S105). Next, the similarity calculation unit 316 calculates the similarity DEG_S between the word vector VCT_P of the quantified operation object ID and each word vector VCT_L of the quantified label information INF_L using the cosine similarity as described above (S106).

[0090] Next, the label information estimation unit 317 estimates that the label information INF_Le having the highest similarity DEG_S to the operation object ID among the similarities DEG_S calculated as described above corresponds to the operation object ID (S107).

[0091] Next, the log information creation unit 318 creates operation log information INF_G including the operation object ID and the estimated label information INF_Le (S108). When creation of the operation log information INF_G is completed, the process returns to step S101. The log information creation unit 318 transmits the created operation log information INF_G to the operation log storage device 5 at any timing.

[0092] 1.3 Effects of this embodiment It is considered that in business system screens, similar names are often used for the IDs of operation components (operation component IDs) and the label information INF_L that explains the operation components to users. For this reason, this embodiment utilizes the similarity between the operation component IDs and the label information INF_L.

[0093] The operation log acquisition device 1 according to this embodiment includes an event detection unit 311, a similarity calculation unit 316, a label information estimation unit 317, and a log information creation unit 318. The event detection unit 311 detects an operation event when a user performs an operation on a screen including a plurality of operation components and a plurality of pieces of label information INF_L, and identifies the operated operation component. The similarity calculation unit 316 calculates a similarity DEG_S between the operation component ID and each piece of label information INF_L. The label information estimation unit 317 estimates that the label information INF_Le having the highest similarity DEG_S with the operation component ID among the calculated similarities DEG_S corresponds to the operation component ID. The log information creation unit 318 creates operation log information INF_G including the operation component ID and the label information INF_Le.

[0094] The operation log acquisition device 1 according to this embodiment further includes a word digitization unit 315. The word digitization unit 315 digitizes each of the operation object ID and the label information INF_L using a distributed representation of a word vector. The similarity calculation unit 316 calculates a similarity DEG_S between the word vector VCT_P of the digitized operation object ID and each word vector VCT_L of the digitized label information INF_L using cosine similarity.

[0095] As a result, information supplementing the user's operation, i.e., label information INF_L corresponding to the operation object ID, is added to the operation log information INF_G. Therefore, the operation object ID and the label information INF_L corresponding to the operation object ID can be identified on the operation log information INF_G. As a result, according to this embodiment, even if the system from which the log was acquired is updated and the ID system is different, the terminal user's operation can be identified. Furthermore, the terminal user's operation can be identified without comparing the operation object ID of the business system from which the log was acquired with the screen data. Furthermore, analysis results that are easy for people to read can be created using only the operation log information INF_G. Materials for explaining the operation procedures to people can be created from the operation log information INF_G.

[0096] 2. Modifications etc. In the flowcharts described in the above embodiments, the order of the processes can be changed as much as possible.

[0097] It should be noted that the present invention is not limited to the above-described embodiments, and various modifications can be made in the implementation stage without departing from the spirit of the invention. Furthermore, the embodiments may be implemented in appropriate combinations, in which case the combined effects can be obtained. Furthermore, the above-described embodiments include various inventions, and various inventions can be extracted by selecting and combining the multiple disclosed constituent elements.

[0098] DESCRIPTION OF SYMBOLS 1... Operation log acquisition device 2... Input unit 3... Processing unit 4... Display unit 5... Operation log storage device 31... Processor 32... Memory 33... Communication interface 311... Event detection unit 312... Screen data acquisition unit 313... Operation information acquisition unit 314... Label information acquisition unit 315... Word digitization unit 316... Similarity calculation unit 317... Label information estimation unit 318... Log information creation unit

Claims

1. An operation log acquisition device comprising: a detection unit that detects an operation event when a user performs an operation on a screen including a plurality of operation parts and a plurality of label information, and identifies a first operation part that has been operated among the operation parts; a calculation unit that calculates a similarity between the ID of the first operation part and each of the label information; an estimation unit that estimates that the first label information having the highest similarity with the ID corresponds to the ID among the calculated similarities; and a creation unit that creates operation log information including the ID and the first label information.

2. The operation log acquisition device according to claim 1, further comprising a quantification unit that quantifies each of the ID and the label information using a distributed representation of word vectors, wherein the calculation by the calculation unit includes calculating a similarity between a first word vector of the quantified ID and each of second word vectors of the quantified label information using cosine similarity.

3. The quantification by the quantification unit includes: after dividing the ID into one or more words, numerically quantifying each of the one or more words obtained by dividing the ID using a distributed representation of word vectors, and synthesizing the word vectors numerically quantified for the ID to numerically quantify the ID as the first word vector; and after dividing each of the label information into one or more words, numerically quantifying each of the one or more words obtained by dividing each of the label information using a distributed representation of word vectors, and synthesizing the word vectors numerically quantified for each of the label information to numerically quantify each of the label information as the second word vector. The operation log acquisition device according to claim 2.

4. An operation log acquisition method comprising: detecting an operation event when a user performs an operation on a screen including a plurality of operation parts and a plurality of label information, and identifying a first operation part that has been operated among the operation parts; calculating a similarity between the ID of the first operation part and each of the label information; estimating that the first label information having the highest similarity with the ID corresponds to the ID among the calculated similarities; and creating operation log information including the ID and the first label information.

Images