Communication during work analysis device and communication during work analysis system

Abstract

To provide a communication analysis device and a communication analysis system capable of evaluating whether communication between a conductor and workers at a work site is appropriately taken.SOLUTION: A communication during work analysis device for analyzing text data of conversation at a work site includes analysis means 12 for specifying whether the text data of an utterance is a conductor's utterance or a worker's utterance, wherein the text data of conversation includes conversation between the conductor and the worker at the work site, and evaluation means 12 for analyzing the utterance of the conductor or the worker to evaluate the communication during work.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a work communication analysis device and a work communication analysis system that analyze and evaluate communication between a leader and workers at a work site. [Background technology]

[0002] 2. Description of the Related Art Conventionally, systems for analyzing and evaluating business communications have been known. For example, Patent Document 1 discloses a system that recognizes the content of speech from voice data spoken by personnel involved in work, converts the speech into text data, obtains data indicating the personnel's work-related actions and data indicating the status of the work, extracts conversation data related to the work from the text data based on the content of the speech recognized by the voice recognition unit, and analyzes the correlation between the conversation data, data indicating the actions, and data indicating the status of the work.

[0003] Furthermore, Patent Document 2 discloses a system that converts on-site voice data into text data, detects keywords from the text data, and determines safety awareness evaluation based on the number of detected keywords.

[0004] Furthermore, Patent Document 3 discloses a meeting confirmation and evaluation system that creates meeting minutes (daily report) from recording files of meetings such as morning meetings held at each work site, confirms, quantifies, and evaluates various matters that must be confirmed at the work site as a matter of course, displays the evaluation results on a monitor, and also makes it possible to output lists, etc. In addition, Patent Document 4 discloses a system in which the names of questions, answers, and voiceprint information obtained from the inspector's own voice are registered in advance, and the inspector's name is saved as inspection data along with the inspection date and time, questions related to the inspection information, and answers. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2019-36205 [Patent Document 2] Japanese Patent Publication No. 2020-197811 [Patent Document 3] Japanese Patent Publication No. 2022-102042 [Patent Document 4] Patent Publication No. 2021-189502 Summary of the Invention [Problem to be solved by the invention]

[0006] At work sites such as power plants, workers generally work under the supervision and instructions of a leader to ensure safety and reliability. Therefore, at such work sites, it is important to evaluate whether the leader and workers are communicating appropriately and working together, such as whether the workers are properly listening to the leader's instructions and performing their tasks, or whether the leader is properly aware of what the workers are doing. However, Patent Documents 1-4 evaluate the work and behavior of workers based on voice data at the work site, and are unable to evaluate whether appropriate communication is taking place between the supervisor and the workers at the work site.

[0007] An object of the present invention is to provide a communication analysis device and a communication analysis system that can evaluate communication between a leader and workers at a work site. [Means for solving the problem]

[0008] The work-time communication analysis device of the present invention is a work-time communication analysis device that analyzes text data of conversations at a work site, the text data including conversations at a work site between a conductor and workers, and is equipped with an analysis means that identifies whether utterances in the text data are utterances of the conductor or the workers, and an evaluation means that analyzes the utterances of the conductor or the workers and evaluates communication at work.

[0009] In the above-mentioned work-time communication analysis device, the analysis means can analyze the text data to identify the speech of the worker that is paired with the speech of the conductor, and the evaluation means can analyze the speech of the paired conductor and worker to evaluate the communication during work. The above-mentioned work-time communication analysis device may further include a database of specific keywords corresponding to a plurality of work types, and an input means for receiving designation of a work type from outside, and the evaluation means may be configured to evaluate the work-time communication by detecting specific keywords corresponding to the work type designated via the input means from the speech of the conductor and / or the speech of the worker. In the above-mentioned work-time communication analysis device, when the specific keyword consists of multiple keywords that should be detected within one conversation, the evaluation means can be configured to evaluate the work-time communication related to the specific keyword as excellent only when all of the multiple keywords are spoken. In the above-mentioned work-time communication analysis device, the evaluation means can be configured to evaluate the work-time communication by detecting whether a first keyword is present in the worker's utterance when it is determined that the worker's utterance is an utterance regarding the worker's voluntary work notice. In the above-mentioned work-time communication analysis device, the evaluation means can be configured to evaluate the work-time communication by detecting whether the conductor's utterance paired with the worker's utterance contains a second keyword corresponding to a first keyword detected in the worker's utterance when it is determined that the worker's utterance is an utterance regarding a voluntary work notice by the worker. In the above-described work-time communication analysis device, the evaluation means can be configured to evaluate the work-time communication by detecting a return or repetition in the conductor's utterance that is paired with the worker's utterance.

[0010] In the above-mentioned work-time communication analysis device, when the evaluation means determines that the conductor's utterance is an utterance regarding work instructions from the conductor, it can be configured to identify the worker's utterance that pairs with the conductor's utterance that includes the specific keyword, and detect a third keyword that corresponds to the specific keyword from the identified worker's utterance, thereby evaluating the work-time communication. In the above-described work-time communication analysis device, the evaluation means can be configured to evaluate the work-time communication by detecting a return or repetition in the utterance of the worker that is paired with the conductor's utterance. In the above-mentioned work-time communication analysis device, the evaluation means can be configured to evaluate the work-time communication by detecting whether the specific keyword is present in the conductor's speech when it is determined that the conductor's speech is about work instructions from the conductor. In the above-described work communication analysis device, the evaluation means may be configured to have a function of outputting the evaluation as a deviation value distribution. The work communication analysis device may further comprise a text generation means for generating the text data from recorded data of conversations at the work site.

[0011] The work communication analysis system of the present invention comprises a plurality of wireless terminals for conversation between a leader and workers, a recording terminal equipped with means for recording conversations at the work site and outputting the recorded data, and the work communication analysis device. [Effects of the Invention]

[0012] According to the present invention, communication between a leader and workers at a work site can be evaluated. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a configuration diagram of a work communication analysis system according to a first embodiment. [Figure 2] FIG. 2 is a diagram showing an example of data items managed in a database according to the first embodiment. [Figure 3] FIG. 10 is a diagram showing an example of an output of an analysis result by an analysis unit. [Figure 4] FIG. 10 is a diagram illustrating a first conversation pattern in which the conductor is the first speaker when performing a unit task. [Figure 5] FIG. 10 is a diagram illustrating a second conversation pattern in which a worker is the first speaker when performing a unit task. [Figure 6] 10 is a flowchart illustrating the processing of the work communication analysis system. [Figure 7] 10 is an example of a report showing the results of an evaluation of communication between a leader and workers during work. [Figure 8] FIG. 10 is a diagram for explaining the evaluation of return and repetition in the first conversation pattern of the second embodiment. [Figure 9] FIG. 10 is a diagram for explaining the evaluation of return and repetition in the second conversation pattern of the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0014] Hereinafter, with reference to the drawings, embodiments of a work communication analysis device and a work communication analysis system according to the present invention will be described. The work communication analysis device according to the present invention analyzes and evaluates whether communication between a supervisor and a worker during work is being carried out appropriately. Here, a supervisor is a person who supervises work and gives instructions to the workers regarding the work. Furthermore, a worker is a person who performs work according to the supervisor's instructions. Note that the supervisor may also perform work. In an actual work site, there may be multiple workers for one supervisor. However, in the embodiment described below, each one-to-one pair of a supervisor and each worker is evaluated. Furthermore, the work site analyzed by the work communication analysis device according to the present invention is not particularly limited as long as the workers work under the direction and supervision of a supervisor. Examples include sites such as power distribution construction, plant startup and shutdown operations, electrical construction, construction work, and other construction sites.

[0015] First Embodiment Fig. 1 is a configuration diagram of a work communication analysis system 100 according to the first embodiment. As shown in Fig. 1, the work communication analysis system 100 according to this embodiment includes a work communication analysis device 1, a conductor wireless terminal 2 used by a conductor, and worker wireless terminals 3a and 3b used by workers.

[0016] The conductor wireless terminal 2 has a communication device (not shown) and is capable of wirelessly communicating with the worker wireless terminals 3a and 3b. The conductor wireless terminal 2 also has a microphone and speaker (not shown) and has a transceiver function of collecting the voice of the conductor with the microphone and transmitting the voice data to the worker wireless terminals 3a and 3b via the communication device, as well as receiving worker voice data from the worker wireless terminals 3a and 3b via the communication device and outputting the worker voice from the speaker based on the received worker voice data.

[0017] The conductor radio terminal 2 also has a recorder 21 for recording voice data. Specifically, the recorder 21 records the voice spoken by the conductor as voice data, and also records the voice data of the workers' voices received from the worker radio terminals 3a and 3b. The recorder 21 also associates the recorded voice data with data on the date and time of the voice speech, making it possible to understand the order in which the worker's voice and the conductor's voice were spoken. The recorder 21 may be configured as an independent device separate from the conductor radio terminal 2.

[0018] The worker wireless terminals 3a and 3b have a communication device (not shown) and are capable of wireless communication with the conductor wireless terminal 2. The worker wireless terminals 3a and 3b also have a microphone and speaker (not shown) and have a transceiver function of collecting the voice of the worker with the microphone and transmitting the voice data to the conductor wireless terminal 2 via the communication device, as well as receiving the conductor's voice data from the conductor wireless terminal 2 via the communication device and outputting the conductor's voice from the speaker based on the received conductor's voice data.

[0019] (Work Communication Analysis Device 1) 1, the work communication analysis device 1 is a computer having an input device 11, a calculation device 12, a database 14, and an output device 15. The work communication analysis device 1 can also communicate with a conductor wireless terminal 2 via wired or wireless means, and can acquire voice data of a conductor and / or workers recorded in a recorder 21 from the conductor wireless terminal 2. Each component of the work communication analysis device 1 will be described below.

[0020] The input device 11 is a keyboard or the like, and can be operated by the user to select the type of work to be analyzed (major items described later). The user uses the input device 11 to select the type of work to be analyzed and clicks an icon or button on the screen to start the analysis, and the computing device 12 analyzes the communication between the supervisor and the workers at the work site.

[0021] The database 14 is a relational database that stores evaluation data in which specific keywords corresponding to each type of work are set in order to evaluate communication between a supervisor and workers at the work site. The evaluation data managed by the database 14 will be described later.

[0022] The output device 15 is a display, a printer, or the like, and presents the evaluation results from the calculation device 12 to the user. Specifically, the evaluation results can be displayed on a display or printed on a printer. The work communication analysis device 1 can also be configured to transmit the evaluation results to an external server or a communication terminal such as a smartphone or tablet without having the output device 15. In this case, the user can check the evaluation results by accessing the external server or using a communication terminal such as a smartphone or tablet.

[0023] (Evaluation data) FIG. 2 is a diagram showing an example of data items managed in the database 14 in this embodiment. As shown in FIG. 2, in this embodiment, work at a work site is categorized and managed into major items, medium items, and small items. The major items indicate the type of work selected by the input device 11, and items indicating a series of work at a work site, such as installing a construction switch, are set. The medium items are set as broad categories of a series of work corresponding to the major items. For example, in the example of FIG. 2, the series of work in the major item "construction switch (installation)" can be categorized into "installation work" and "edge disconnection work" for the construction switch, so "installation work" and "edge disconnection work" are set as medium items. Furthermore, the small items are set as unit work that make up the work of the medium item. For example, in the example shown in FIG. 2, "installation work" for the construction switch consists of "hanging and lowering heavy objects" and "indirect live-line work (cable connection)," and these unit work are set as small items.

[0024] The evaluation data items managed in the database 14 include specific keywords. The specific keywords are keywords that indicate the target object (the equipment or object to be worked on) or target action (the action to be worked on) of each unit task defined in the subitems of FIG. 2 . In this embodiment, for each unit task in the subitems, keyword evaluation is performed to evaluate whether the communication between the leader and the worker in the task is good by determining whether the specific keyword set for that task is uttered. In this embodiment, as shown in FIG. 2 , the specific keywords are divided into multiple groups, such as first specific keywords and second specific keywords. In the example shown in FIG. 2 , there are unit tasks in which the communication between the leader and the worker is evaluated by determining whether both the first specific keyword and the second specific keyword are uttered, as well as unit tasks in which only the first specific keyword is set. Note that in the example shown in FIG. 2 , first specific keywords and second specific keywords are set as groups of specific keywords, but the number of specific keyword groups may be three or more.

[0025] The data items (major items, medium items, and minor items) and specific keywords managed in the database 14 can be set in advance by an administrator or maintenance staff member, and can also be modified as needed.

[0026] Furthermore, although not shown, in this embodiment, the database 14 can be configured to store a wordbook in order to deal with variations in words. For example, if expressions such as "cut" and "open" are used with the same meaning for the word "open" in the same task, by associating the word "open" with the words "cut" and "open" in the wordbook, it becomes possible to determine that a word equivalent to "open" has been uttered when the words "cut" and "open" are used during evaluation.

[0027] In addition, in FIG. 2, information indicating the order of the work is attached to each task in the minor items. The evaluation means described below evaluates whether each task is performed in order with appropriate communication based on the order attached to each task in the minor items. In the example shown in FIG. 2, the work of the major item "Construction switch (installation)" is composed of the medium items "1. Installation work" and "2. Edge cutting work," and the order of the work is set in order from top to bottom as follows: the minor items "1-1. Lifting and lowering heavy objects" and "1-2. Indirect live-line work (cable connection)" that constitute the medium item "1. Installation work," and the minor items "2-1. Turning on the construction switch" and "2-2. Indirect live-line work (cutting)" that constitute the medium item "2. Edge cutting work."

[0028] (Conversation Evaluation Program P) The arithmetic device 12 executes the conversation evaluation program P stored in the storage device 13 to analyze text data of conversations between a conductor and workers at a work site and evaluate the communication between the conductor and workers during work. To perform the above evaluation, the arithmetic device 12 has a text data generation means for acquiring text data of conversations between the conductor and workers during work, an analysis means for identifying utterances by the conductor and utterances by the workers from the text data, an evaluation means for evaluating the communication between the conductor and the workers, and an output means for outputting the evaluation results. Note that the conversation evaluation program P for executing the above means does not need to be stored in the built-in storage device 13, and the work communication analysis device 1 may access an external server and execute the conversation evaluation program P on the server.

[0029] (Text data generation means) The text data generation means converts voice data of conversations between a leader and workers at a work site into text data. In this embodiment, the text data generation means acquires voice data of the conductor and workers recorded in the recorder 21 of the conductor wireless terminal 2 used by the conductor, and converts it into text data. Here, the voice data acquired from the conductor wireless terminal 2 may be recorded in association with information such as a file name, team name, work content, and work date and time. In this embodiment, in order to improve the conversion accuracy when converting voice data into text data, a voice recognition program that converts voice data into text data is used. As the voice recognition program, it is preferable to use a program that uses AI (Artificial Intelligence) that has been trained using speech data actually used at the work site as training data (hereinafter referred to as an "AI program").

[0030] (analytical means) The analysis means determines whether each utterance is a conductor's utterance or a worker's utterance based on the text data generated by the text data generation means. Specifically, the analysis means analyzes the context of the text data using natural language processing to determine whether each utterance is a conductor's utterance or a worker's utterance. In this embodiment, an AI program determines whether the utterance is a conductor's utterance or a worker's utterance based on the highest probability. An example of an AI program using such natural language processing technology is the BERT model. The analysis means also has the function of analyzing the content of each utterance using natural language processing and inferring the corresponding task (the sub-item task in Figure 2). The analysis means labels each utterance with a worker / conductor label, outputs the analysis results linked to the corresponding task in the form of table data, and links this to the evaluation means. Figure 3 shows an example of the analysis results output by the analysis means. The analysis means predicts the corresponding unit task from the underlined and double-underlined keywords and links them to the utterance.

[0031] (Evaluation method) The evaluation means evaluates whether communication between the conductor and the workers during the work was appropriate based on the utterances of the conductor and the workers identified by the analysis means. The evaluation means performs at least a) single-task keyword evaluation, b) multiple-task keyword evaluation, and c) review / repetition evaluation.

[0032] a. Single-task keyword evaluation The single task keyword evaluation evaluates the communication between the conductor and the workers during the work by referring to the database 14 and detecting specific keywords corresponding to each unit task of the work type (major items in Figure 2) specified via the input device 11 from the conductor's speech and / or the workers' speech.

[0033] Specifically, whether the conductor or worker communicates the object to be worked on (target object) and the action to be performed in the work (target action) to the other party in work instructions from the conductor or in a voluntary work notice from the worker is evaluated based on whether or not keywords are uttered. In the first embodiment, as shown in FIG. 2, a specific keyword corresponding to each work is set for each work. The evaluation means evaluates whether appropriate communication is taking place by determining whether the conductor or worker utters a specific keyword during conversation. For example, in the example shown in FIG. 2, for the work subitem "1-1. Lifting and lowering heavy objects," specific keywords such as "lifting" and "raising," which indicate the target action, are set. If the conductor who instructed the work utters a specific keyword such as "lifting" or "raising" when performing the work subitem "1-1. Lifting and lowering heavy objects," the evaluation means can evaluate the keyword evaluation for the work subitem "1-1. Lifting and lowering heavy objects" as "good." In addition, if the worker who announced the work makes a speech containing a specific keyword during the work, and the conductor responds that he or she agrees to the work, the keyword evaluation can be evaluated as "Good."

[0034] In some cases, the single task keyword evaluation is evaluated as "good" when both the first and second specific keywords are detected. For some unit tasks, a first specific keyword indicating the target object of the task and a second specific keyword indicating the target action of the task are set, and the evaluation means evaluates the keyword evaluation as "good" only when both the first and second specific keywords are uttered. For example, in the example shown in FIG. 2, for the subitem "1-2. Indirect live-line work (cable connection)," first specific keywords indicating the target object of the task are set as "red" and "red phase," and second specific keywords indicating the target action of the task are set as "connect," "join," and "edge connection." Therefore, for the task of the subitem "1-2. Indirect live-line work (cable connection)," the evaluation means can evaluate the keyword evaluation for the subitem as "good" only when both the first and second specific keywords are detected in the utterance.

[0035] b. Multitasking Keyword Evaluation Multi-task keyword evaluation is performed based on whether a keyword appears in each of multiple related unit tasks. For example, in Figure 2, if the specific keyword for unit task "1-1" is not uttered before the specific keyword for unit task "1-2" is detected, the evaluation of unit task "1-1" is determined to be "X." Note that unit tasks "1-A" and "1-B" are subject to the single task keyword evaluation of "1. Installation Work," and therefore, if the specific keyword is uttered before the specific keyword is detected in "2-1," the single task keyword evaluation will be "○," regardless of the order in which the utterances were detected.

[0036] c.Report and repetition evaluation The report / readback evaluation is evaluated as "Good" only if the first speaker utters the three elements of the work object, the target action, and an indication of understanding. Here, the work object and target action may or may not be set as specific keywords. In the example of the unit task "1-2. Indirect live-line work (cable connection)" in Figure 2, "red, red phase, ...", which indicates the red phase cable that is the work object, is set as the first specific keyword, and the target action "connect, splice, edge connection, ..." is set as the second specific keyword. However, in the unit task "1-1. Lifting and lowering heavy objects", only the target action "lifting, lifting, ..." is set as the first specific keyword.

[0037] d. Additional evaluation The evaluation means can incorporate a known AI program to analyze the voice data and determine "emotions (anger, impatience)." For example, if the voice data contains speech at a slower speed, trembling, or slips of the tongue, it can be determined that impatience is present. Also, a function may be provided to evaluate the number of times the conductor and workers speak as the total number of conversations. It is also possible to incorporate a known AI program to analyze audio data and determine "clarity." In addition, a function for storing the history of past evaluation results of each work group in the database 14 may be provided.

[0038] (Judgment by conversation pattern) In this embodiment, a conversation pattern is determined prior to each evaluation. In a speech pair when performing the same unit work, even if the content of the speech between the first speaker and the second speaker is the same, different evaluations may be made depending on the conversation pattern. The speech pair is classified into either the first conversation pattern or the second conversation pattern. Note that the speech pair is evaluated based on the conductor's speech and the worker's speech (in no particular order), and the speech between the workers is not evaluated.

[0039] The first conversation pattern is a conversation pattern in which the conductor gives work instructions and the worker reports and repeats them. FIG. 4 is a diagram explaining the first conversation pattern in which the conductor is the first speaker when performing a unit task, and is an example of a conversation when performing the unit task "1-2. Indirect live-line work (cable connection)" shown in FIG. 2. When performing the unit task "1-2. Indirect live-line work (cable connection)," if the conductor gives the work instruction "Connect the red-phase cable," the evaluation means refers to the database 14 and determines from the text data of the conductor's utterance that the first specific keyword "red-phase" and the second specific keyword "connect" were uttered, and evaluates the keyword evaluation of the conductor's utterance as "good." At this time, the worker's response is not taken into consideration. On the other hand, for the evaluation of the report and repetition, the evaluation will be "○" only if the subject of the task, "red phase," and the target action, "connect," are spoken, and "understood" is spoken to indicate understanding.

[0040] The second conversation pattern is one in which the worker spontaneously gives advance notice of the work and the commander reports and repeats it. Figure 5 is a diagram explaining the second conversation pattern in which the worker is the first speaker when performing a unit task, and, like Figure 4, is an example of a conversation when performing the unit task "1-2. Indirect live-line work (cable connection)." When performing the unit task "1-2. Indirect live-line work (cable connection)," if a worker gives a work notice saying, "I'll connect the red-phase cable," the evaluation means references the database 14 and detects from the text data of the worker's speech that the first specific keyword "red-phase" and the second specific keyword "connect" were uttered. Here, in the second conversation pattern, the keyword evaluation is performed taking into consideration not only the worker's speech but also the conductor's response. If the report / repetition evaluation is "○," the keyword evaluation is also "○." If the conductor utters "understood," indicating understanding, the keyword evaluation is also "○." This is because it can be confirmed that there are no errors in the worker's work notice. On the other hand, if there is no response from the conductor, it is impossible to determine whether the work notice is correct, and the keyword evaluation is "×." Similarly, if the conductor responds by encouraging the worker to correct the work content, the keyword evaluation is also "×" because the conductor's expression of understanding cannot be confirmed.

[0041] (Method for determining conversation patterns) Conversation patterns are determined based on the vocabulary contained in the speech. First, morphological analysis is performed on each utterance, and sentence endings are obtained from the divisions of the obtained Japanese sentences and classified into three types of utterance types (work instructions, work notices, and report / repetition). In this embodiment, utterance types and sentence ending expressions are associated and stored in the database 14, and the evaluation means refers to the database 14 to determine the utterance type according to the obtained sentence ending expression. For example, in the utterance "I understand that the red-phase cable is connected," the sentence ending "I understand" is obtained from the results of morphological analysis. "I understand" is treated as "report / repetition." Next, the evaluation means determines a conversation pattern based on the determined type of speech. The first conversation pattern is determined to be the relevant conversation pattern when the conductor's speech is classified as a work instruction and the next worker's speech is a report or repetition. The second conversation pattern is determined to be the relevant conversation pattern when the worker's speech is classified as a work notice and the next conductor's speech is classified as a work instruction or a report or repetition.

[0042] (Processing flow) FIG. 6 is a flowchart illustrating the processing of the work communication analysis system 100. STEP 601: When using this system 100, it is necessary to record the conversation between the leader and the worker during work. After the work is completed, the voice data of the recorded conversation is used to perform analysis by the work communication analysis device 1. STEP 602: The task to be analyzed is identified, the corresponding task type is selected, and the corresponding voice data is imported into the conversation evaluation program P to start the analysis. The conversation evaluation program P performs the following analysis using the evaluation data corresponding to the selected task type. STEP 603: The conversation evaluation program P executes the text data generating means to convert the input voice data into text data. STEP 604: The conversation evaluation program P executes the analysis means to analyze the acquired text data and identify whether the speaker of each utterance is a conductor or a worker. STEP 605: The conversation evaluation program P executes the analysis means to analyze the acquired text data and estimate utterances corresponding to unit tasks that constitute the target task. STEP 606: The conversation evaluation program P executes the evaluation means to select utterance pairs to be evaluated. In principle, utterance pairs are detected and selected from the earliest to the latest time slots in the time series. STEP 607: The conversation evaluation program P uses the evaluation means to determine whether the selected utterance pair corresponds to the first conversation pattern or the second conversation pattern, and stores the evaluation result in the storage device 13. STEP 608: If the selected utterance pair is of the first conversation pattern, the conversation evaluation program P uses the evaluation means to perform a keyword evaluation of the selected utterance pair, and stores the evaluation result in the storage device 13. If the selected utterance pair is of the second conversation pattern, the conversation evaluation program P performs the review and repetition evaluation in STEP 609 first, and then performs the keyword evaluation. STEP 609: The conversation evaluation program P uses the evaluation means to perform a read-back / recapitulation evaluation of the selected utterance pair, and stores the evaluation result in the storage device 13. STEP 610: The conversation evaluation program P executes STEPs 606 to 609 until evaluation of all active utterances is completed. STEP 611: When the conversation evaluation program P has finished evaluating all the utterances currently being evaluated, it causes the output means to output the evaluation results to the output device 15.

[0043] (output means) The output means can output the evaluation results as a graphical report. FIG. 7 shows an example of a report output by the output means as a result of evaluation of communication between a leader and workers during work. In the pentagonal graph shown in the upper right of Fig. 7, the evaluation results of each work team are output as absolute evaluations. For example, the output means can be configured to count an evaluation of "o" as 1 point for each task (sub-item) included in the task type (major item) selected via the input device 11, and output the evaluation results as scores with full points when all tasks are "o."

[0044] Additionally, in the middle section of Figure 7, the evaluation of each work team is output as a deviation value relative to the overall evaluation. Here, the keyword evaluation is displayed as "direct command and monitoring implementation status," and the feedback and replay evaluation is displayed as "replay and replay implementation status." Outputting the deviation value makes the position of each work team clear in the overall evaluation, and is expected to prevent a decline in motivation. Although not shown in Figure 7, a function can be provided to automatically generate comments such as areas for improvement and points to be aware of based on the evaluation results. 7, a graph showing the progress of the evaluation results based on the evaluation history of the work team stored in the database 14 is displayed.

[0045] As described above, the work-time communication analysis device 1 according to the first embodiment can appropriately evaluate whether the commander and workers were communicating appropriately while working on the task by evaluating whether the specific keywords set for the unit task were spoken. Furthermore, by performing a multiple task keyword evaluation, it is possible to evaluate whether a plurality of related unit tasks are being performed in an appropriate order. Furthermore, by conducting a feedback and repetition evaluation, it is possible to evaluate whether the work is being carried out appropriately while the feedback and repetition are being carried out.

[0046] Second Embodiment Next, a second embodiment of the present invention will be described. The work communication analysis system 100 and work communication analysis device 1 according to the second embodiment have the same configuration as the work communication analysis system 100 and work communication analysis device 1 according to the first embodiment, and differ from the first embodiment in that the review and repetition evaluation is a three-level evaluation of "○△×" using a conversion vocabulary book, but the other configurations and operations are the same. Below, the differences will be described, and a description of the similarities will be omitted.

[0047] The processing flow in the second embodiment is the same as the processing flow in the first embodiment shown in Fig. 6 in terms of the step level to be performed. Also in the second embodiment, the report / repetition evaluation is evaluated as "○" only when the three elements of the work object, the target action, and the expression of understanding are uttered. On the other hand, in the second embodiment, even when the second speaker utters another word that corresponds to the keyword uttered by the first speaker, word conversion is performed using the conversion vocabulary book to match it with the keyword uttered by the first speaker. In addition, in the second embodiment, if utterance of two of the above three elements is detected, the evaluation is given as "△", and if only one or less elements are detected, the evaluation is given as "×".

[0048] Fig. 8 is a diagram for explaining the return / repetition evaluation in the first conversation pattern of the second embodiment, and Fig. 9 is a diagram for explaining the return / repetition evaluation in the second conversation pattern of the second embodiment. In the reply examples marked with an asterisk in the ○ evaluation rows of Fig. 8 and Fig. 9, it has been determined that the above three elements have been uttered by performing word conversion using the conversion wordbook.

[0049] According to the work-time communication analysis device 1 of the second embodiment described above, the review and repetition evaluation is performed using a conversion vocabulary book and is graded on a three-level scale of "○△×", making it possible to evaluate work-time communication in more detail.

[0050] Although the preferred embodiments of the present invention have been described above, the technical scope of the present invention is not limited to the above-described embodiments. Various modifications and improvements can be made to the above-described embodiments, and such modifications and improvements are also included in the technical scope of the present invention.

[0051] For example, in the above-described embodiment, a configuration has been exemplified in which the work communication analysis device 1 acquires voice data from the conductor wireless terminal 2 and converts the acquired voice data into text data to acquire text data of the speech of the conductor and / or worker, but the configuration is not limited to this, and the voice data spoken by the conductor and / or worker can also be acquired from another device (another computer or server).Furthermore, in the above-described embodiment, a configuration has been exemplified in which the work communication analysis device 1 acquires voice data from the conductor wireless terminal 2, but a configuration in which pre-converted text data is acquired from an external device instead of voice data can also be used.

[0052] In the above-described embodiment, the work communication analysis device 1 is configured to identify whether each utterance is the conductor's or worker's utterance from the context based on the text data of the conversation between the conductor and the worker. However, differently, for example, the conductor wireless terminal 2 can add information indicating that the conductor's voice is the conductor's voice to the voice data collected by the conductor wireless terminal 2, and the conductor wireless terminal 2 can add information indicating that the voice data received from the worker wireless terminals 3 a and 3 b is the worker's voice to identify whether the utterance is the conductor's or worker's voice without analyzing the context of the text data of the conversation. Furthermore, by using voice data in addition to text data, it is also possible to identify utterances made by the same person from the voiceprint contained in the voice data. [Explanation of symbols]

[0053] 100...Work-related communication analysis system 1. Work-related communication analysis device 11...Input device 12...Arithmetic device 13...Storage device 14...Database 15...Output device 2...Commander radio terminal 21...Recorder 3a, 3b...Worker wireless terminal

Claims

1. A work-time communication analysis device that analyzes text data of conversations at a work site, the text data includes a conversation between a leader and a worker at a work site; an analysis means for identifying whether the speech in the text data is a speech of a conductor or a speech of a worker; and an evaluation means for analyzing the utterances of the conductor or the worker to evaluate the communication during work.

2. The analysis means can analyze the text data to identify an utterance of a worker that is paired with an utterance of the conductor, 2. The work communication analysis device according to claim 1, wherein the evaluation means is capable of evaluating communication during work by analyzing utterances of the conductor and the worker who are paired.

3. Furthermore, the system is provided with a database of specific keywords corresponding to a plurality of work types and an input means for receiving designation of the work type from an external source, The work-time communication analysis device according to claim 1, wherein the evaluation means evaluates communication during work by detecting specific keywords corresponding to the type of work specified via the input means from the conductor's speech and / or the worker's speech.

4. When the specific keyword is composed of multiple keywords to be detected in one conversation, 4. The work-time communication analysis device according to claim 3, wherein the evaluation means evaluates the work-time communication related to a specific keyword as excellent only when all of the plurality of keywords are uttered.

5. The work-time communication analysis device of claim 3, wherein the evaluation means, when determining that the worker's speech is a speech about the worker's voluntary work notice, evaluates communication during work by detecting whether a first keyword is present in the worker's speech.

6. The work-time communication analysis device of claim 3, wherein the evaluation means, when determining that the worker's speech is a voluntary work notice by the worker, evaluates communication during work by detecting whether the conductor's speech paired with the worker's speech contains a second keyword corresponding to a first keyword detected in the worker's speech.

7. The work-time communication analysis device according to claim 6 , wherein the evaluation means evaluates the work-time communication by detecting a return or repetition in the conductor's utterance that is paired with the worker's utterance.

8. The work-time communication analysis device according to claim 3, wherein the evaluation means, when determining that the conductor's speech is a speech about work instructions from the conductor, identifies the worker's speech that pairs with the conductor's speech that includes the specific keyword, and detects a third keyword that corresponds to the specific keyword from the identified worker's speech, thereby evaluating communication during work.

9. The work-time communication analysis device according to claim 8 , wherein the evaluation means evaluates the work-time communication by detecting a return or repetition in the worker's utterance that is paired with the conductor's utterance.

10. The work-time communication analysis device according to claim 3, wherein the evaluation means, when determining that the conductor's speech is a speech about work instructions by the conductor, evaluates communication during work by detecting whether the specific keyword is present in the conductor's speech.

11. The work-time communication analysis device according to claim 1 , wherein the evaluation means has a function of outputting the evaluation as a deviation value distribution.

12. 12. The work communication analysis device according to claim 1, further comprising a text generation unit that generates the text data from recorded data of a conversation at a work site.

13. a plurality of wireless terminals for communication between the conductor and the workers; a recording terminal having means for recording conversations at a work site and outputting the recorded data; A work communication analysis system comprising: the work communication analysis device according to claim 12.

Images