Conversation analysis system and method

Abstract

The invention provides a dialogue analysis system and method, and the system comprises a data processing module which is used for carrying out the semantic segmentation of data collected in real time, and obtaining a plurality of semantic units; wherein the data is related to conversations among a plurality of learners in the cooperative learning process; the dialogue analysis module is used for comparing a dialogue sequence with a preset dialogue mode in an expert rule base to obtain a dialogue analysis result after the dialogue sequence is formed based on the semantic units; the feedback module is used for generating feedback information based on the dialogue analysis result; wherein the feedback information is used for feeding back problems existing in the dialogue. According to the method and the device, the dialogues among a plurality of learners in the cooperative learning process can be analyzed in real time, the problems existing in the dialogues are fed back, the real-time performance of dynamic analysis is high, the recognition and adaptive capacity of dynamic interaction is enhanced, and the availability of diversified learning situations is improved.

Description

Technical Field

[0001] The present invention relates to the field of natural language processing, and in particular to a conversation analysis system and method. Background Art

[0002] At present, collaborative learning is an important educational method and is widely used in various educational and training situations. Its core lies in completing tasks and solving problems through interaction and cooperation between students or team members. This teaching method can not only improve individual academic performance, but also cultivate teamwork ability, problem-solving ability and social skills. Therefore, the effectiveness of collaborative learning depends on the quality of interaction between participants, which involves not only the sharing of knowledge and skills, but also emotional support and teamwork.

[0003] In practice, the effectiveness of collaborative learning depends mainly on the quality of the dialogue. However, related research has mainly focused on analyzing static dialogues, and has failed to fully explore their dynamic sequence and the performance of serialization features in different contexts. The dialogue analysis process for collaborative learning needs to be optimized. Summary of the invention

[0004] In view of this, this specification provides a conversation analysis system and method.

[0005] According to a first aspect of an embodiment of this specification, a conversation analysis system is provided, the system comprising:

[0006] A data processing module is used to semantically segment the data collected in real time to obtain multiple semantic units; wherein the data is related to the dialogue between multiple learners in the collaborative learning process;

[0007] A dialogue analysis module, configured to, after forming a dialogue sequence based on the plurality of semantic units, compare the dialogue sequence with a preset dialogue pattern in an expert rule library to obtain a dialogue analysis result;

[0008] A feedback module is used to generate feedback information based on the dialogue analysis results; wherein the feedback information is used to feedback problems existing in the dialogue.

[0009] According to a second aspect of an embodiment of this specification, a conversation analysis method is provided, the method comprising:

[0010] Semantically segmenting the data collected in real time to obtain a plurality of semantic units; wherein the data is related to the dialogue between a plurality of learners in the collaborative learning process;

[0011] After a dialogue sequence is formed based on the multiple semantic units, the dialogue sequence is compared with a preset dialogue pattern in an expert rule library to obtain a dialogue analysis result;

[0012] Based on the dialogue analysis result, feedback information is generated; wherein the feedback information is used to feedback problems existing in the dialogue.

[0013] According to a third aspect of the embodiments of this specification, there is provided a conversation analysis device, the device comprising: one or more processors; wherein the processor is used to execute the steps of the method as described in any one of the second aspects.

[0014] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided, on which computer instructions are stored, and when the instructions are executed by a processor, the steps of the method described in any one of the second aspects are implemented.

[0015] According to a fifth aspect of the embodiments of this specification, a computer program product is provided, comprising a computer program / instruction, which, when executed by a processor, implements the steps of the method described in any one of the second aspects.

[0016] The technical solutions provided by the embodiments of this specification may have the following beneficial effects:

[0017] In the embodiments of this specification, the conversations between multiple learners during the collaborative learning process can be analyzed in real time, and feedback can be given on problems existing in the conversations. The dynamic analysis is highly real-time, which enhances the recognition and adaptability of dynamic interactions and improves the usability for diverse learning situations.

[0018] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the specification and, together with the description, serve to explain the principles of the specification.

[0020] Figure 1 This is one of the structural diagrams of a conversation analysis system provided by an exemplary embodiment.

[0021] Figure 2 This is the second structural diagram of a conversation analysis system provided by an exemplary embodiment.

[0022] Figure 3 This is the third structural diagram of a conversation analysis system provided by an exemplary embodiment.

[0023] Figure 4This is one of the flowcharts of a conversation analysis method provided by an exemplary embodiment.

[0024] Figure 5 This is the second flowchart of a conversation analysis method provided by an exemplary embodiment.

[0025] Figure 6 This is the third flowchart of a conversation analysis method provided by an exemplary embodiment.

[0026] Figure 7 This is the fourth flowchart of a conversation analysis method provided by an exemplary embodiment.

[0027] Figure 8 This is the fourth structural diagram of a conversation analysis system provided by an exemplary embodiment.

[0028] Fig. 9 It is a structural diagram of a server provided by an exemplary embodiment. DETAILED DESCRIPTION

[0029] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this specification. Instead, they are merely examples of devices and methods consistent with some aspects of this specification as detailed in the appended claims.

[0030] The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit this specification. The singular forms "a", "the" and "the" used in this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0031] It should be understood that although the terms first, second, third, etc. may be used in this specification to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this specification, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0032] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0033] In related technologies, the challenges of research and practice are mainly reflected in the following three aspects:

[0034] On the one hand, the diversity of relevant coding systems makes it difficult to systematize the analysis of conversations, which affects the uniformity of research results and the generalizability of practice.

[0035] On the other hand, although various coding systems have been widely discussed in the literature, there is still a lack of research on dialogue sequences. The order of dialogue is the key to understanding the dynamic process of cooperative learning, but research in this area has not yet been in-depth.

[0036] On the other hand, relevant technologies are mainly based on post-analysis, which makes it difficult to achieve real-time monitoring and intervention in the collaborative learning process, and thus cannot effectively promote efficient collaboration among learners.

[0037] In the disclosed embodiments, a learning dialogue model based on retrospective data analysis is provided. It can rely on pre-recorded or recorded collaborative learning situations (such as students' dialogue logs, videos or interaction records), and in the subsequent stage, these records are processed offline and coded and classified through a series of analysis tools and models. It aims to extract representative dialogue features from these retrospective data and understand learners' interaction patterns, cognitive processes and social emotional expressions through these features.

[0038] The learning dialogue model based on retrospective data analysis can include the following steps when conducting dialogue analysis:

[0039] Step 1, data collection.

[0040] In the collaborative learning process, students’ conversation data can be collected through audio, video or text recording. The types of data include text conversations, voice conversations, video recordings, log activities, etc. This process does not provide real-time analysis and feedback on students’ conversations.

[0041] Step 2: data preprocessing.

[0042] The collected raw data is cleaned and preprocessed to remove noise, irrelevant information or redundant data. This step usually uses natural language processing (NLP) technology to perform word segmentation, annotation and semantic analysis on text dialogues; while video data may use computer vision technology for dialogue detection.

[0043] Step 3: Coding and classification.

[0044] The processed data are classified into dialogue labels through preset coding frameworks, such as cognitive activity coding, social interaction coding, etc. For example, by defining different coding systems, students' dialogue content is marked, such as questioning, explanation, negotiation, etc., or dialogues are divided into cooperation, competition, emotional support, etc.

[0045] Step 4: Retrospective analysis.

[0046] Based on the encoded data, a series of machine learning algorithms and statistical tools can be used to analyze the characteristics of learners' conversations. This may include sequence analysis, such as analyzing the transition probability and dependency between different turns through the Markov chain model, or analyzing the interaction structure and information flow between learners through social network analysis.

[0047] It is understandable that the above process can be mainly applied to the later reflection and summary analysis of collaborative learning. Its typical usage scenarios include but are not limited to at least one of the following:

[0048] Evaluation of the effectiveness of cooperative learning: After the teaching experiment, the researchers conducted a retrospective analysis to evaluate whether the students showed the expected interactive dialogue (such as effective discussion, collaborative problem solving, etc.) during the learning process, and compared the differences in interaction between different groups.

[0049] Teacher teaching reflection: Teachers can analyze students’ conversation records to understand students’ cognitive conflicts, emotional changes, and team interactions during the collaborative learning process, and then adjust their teaching strategies.

[0050] Student self-reflection: Students reflect on their own participation and cooperation strategies by analyzing the interaction patterns and personal dialogue characteristics of system feedback, and find ways to improve.

[0051] However, the above process also has the following major disadvantages:

[0052] First, there is a lack of real-time analysis and feedback.

[0053] The above solutions mainly rely on post-process data collection and offline analysis, which results in the inability to identify and adjust learners' dialogue patterns in real time during the actual collaborative learning process. This post-processing approach lags behind in the face of dynamically changing learning scenarios, making it difficult to promptly identify and correct bad interactive dialogue patterns, limiting immediate guidance for learners.

[0054] Second, the ability to recognize dynamic interactions is limited.

[0055] The above scheme relies on a preset coding system and a fixed dialogue mode, and mainly analyzes learners' dialogue characteristics through static labels. This method is difficult to capture dynamic and complex interaction changes, especially in the changing cooperative learning situation. This static coding system is too rigid to identify flexible and changing dialogue sequences and situation changes.

[0056] Again, it is difficult to adapt to the complexity of conversations in diverse learning situations.

[0057] The above solutions often rely on specific coding systems or dialogue models to analyze the collaborative learning process, and these models may not fully cover the real interaction of students under different cultural backgrounds, task situations and learning goals. This makes technology A have certain limitations and insufficient adaptability when dealing with diverse learning situations.

[0058] In order to solve the above problems, the embodiments of the present disclosure provide the following dialogue analysis system and solution, which can accurately evaluate the cooperative learning ability of participants. At the same time, it also has a real-time intervention function, which can provide instant interruption support or dialogue suggestions based on the identified dialogue mode during the cooperative interaction process to guide efficient cooperation.

[0059] Among them, the system and method provided by the embodiments of the present disclosure can be applied to a variety of education and training scenarios.

[0060] For example, in classroom teaching, it is possible to identify the interaction patterns of efficient cooperative groups by monitoring students' conversations, and detect and intervene in possible cooperative learning difficulties in real time, thereby helping teachers understand students' cooperative status in a timely manner and provide targeted guidance and intervention.

[0061] For example, in corporate training and team projects, it can help team members identify potential problems in cooperation, promote communication and task collaboration within the team, thereby improving work efficiency and the quality of task completion.

[0062] For example, in cross-cultural collaborative learning, it can help identify communication barriers caused by cultural differences.

[0063] It is understandable that the system and method provided in this specification can also be applied to other scenarios, including but not limited to psychological counseling, medical counseling, social media analysis, customer service, etc. The present disclosure does not limit the application scenarios.

[0064] The following is an introduction to the conversation analysis system provided by the present disclosure.

[0065] Figure 1 FIG. 1 is a schematic diagram of a conversation analysis system provided by an exemplary embodiment. Figure 1 As shown, the system may include:

[0066] Data processing module 101;

[0067] Conversation analysis module 102;

[0068] Feedback module 103.

[0069] In some embodiments, the data processing module 101 may perform semantic segmentation on the data collected in real time to obtain multiple semantic units.

[0070] In an example, the format of the data may include but is not limited to at least one of the following: audio format; video format; text format.

[0071] In one example, the data may be data related to a conversation between multiple learners in a collaborative learning process.

[0072] In one example, the data processing module may include a preprocessing submodule, wherein the preprocessing submodule may be used to perform preprocessing on the above data.

[0073] Exemplarily, the preprocessing includes but is not limited to at least one of the following: noise reduction processing; content segmentation processing; cleaning processing.

[0074] For the audio content in the data in audio format and / or the data in video format, noise reduction processing may be performed to reduce the noise therein.

[0075] Among them, content segmentation processing can be performed on data in video format and / or data in audio format to facilitate subsequent encoding and dialogue analysis.

[0076] Exemplarily, the content segmentation process may be performed in the following manners, but not limited to:

[0077] In one approach, content segmentation may be performed based on conversation turns.

[0078] In the collaborative learning process, each learner's speech (i.e., "turn") can be used as a segmentation unit, which is convenient for direct correspondence to the semantic unit of the dialogue and facilitates subsequent encoding and analysis.

[0079] Alternatively, content segmentation may be performed based on discussion topics.

[0080] During the discussion, learners may switch discussion topics (the same dialogue may include more than one interlocutor), so the content of different tasks needs to be segmented in advance.

[0081] Another way is to perform content segmentation based on task steps.

[0082] In the collaborative learning process, content segmentation is performed based on different steps or stages of the same collaborative task. For example, the part of discussing the plan at the beginning of the task, the part of executing the task, and the part of summarizing the task after the end can all be marked as different segments.

[0083] Among them, cleaning processing can be performed on data in audio format, data in video format and / or data in text format, and the cleaning processing can be used to remove redundant information in the data.

[0084] The redundant information includes but is not limited to at least one of the following: non-semantic filler words; repeated content; environmental noise; blanks and long pauses.

[0085] Among them, non-semantic filler words can refer to some words that have no semantics in the conversation, such as common catchphrases such as "um", "ah", and "that" in the conversation. Although they may be unintentional in actual interactions, they usually do not directly help in understanding the content of the conversation and can be removed.

[0086] Among them, repeated content can refer to the content with the same semantics caused by the incoherence of oral expression in the dialogue. For example, learners may repeat certain words or phrases. These repeated contents will be removed to simplify the dialogue content.

[0087] Among them, environmental noise can refer to background noise appearing in audio and / or video, such as environmental sounds, other people's voices, air conditioning noise, etc., and noise reduction processing can be performed to retain clear voice or visual content.

[0088] Among them, blanks and long pauses can refer to long silent pauses between sentences and words in the dialogue, which can be removed, and such information will not be regarded as meaningful dialogue units. Of course, they can also be regarded as "silent" state instead of redundant information. In the "silent" state, dialogue analysis can be omitted.

[0089] In one example, the data processing module may include a dialog switching submodule, and the dialog switching submodule is used to perform semantic segmentation on the above data to obtain multiple semantic units.

[0090] Exemplarily, the dialogue switching submodule may utilize natural language processing technology to perform semantic segmentation on the above data.

[0091] Exemplarily, semantic segmentation is performed to identify the purpose of a learner's speech.

[0092] Exemplarily, a semantic unit may refer to a language unit that has independent semantics and can be analyzed.

[0093] Among them, the name of the semantic unit is not limited and can be interchanged with "expression unit", "language unit", etc.

[0094] For example, when a conversation facilitator initiates a discussion, he or she may first say, "Let's start the discussion. Today's topic is..." After semantic segmentation, the above language will be regarded as "goal setting" or "task start". For another example, the conversation facilitator may also say, "We can consider the time, location, activities, etc. of the organization event..." After semantic segmentation, the above language will be regarded as "formulating and adjusting strategies".

[0095] In some embodiments, the conversation analysis module 102 may be used to systematically identify and analyze the learner's conversation model through a large language model and intelligent algorithms to obtain conversation analysis results.

[0096] In one example, the conversation analysis module may include a conversation encoding submodule, and the conversation encoding submodule may be used to perform encoding processing on each semantic unit to obtain an encoded semantic unit.

[0097] Exemplarily, the dialogue encoding submodule may encode each semantic unit based on a pre-summarized collaborative learning dialogue encoding framework, thereby obtaining an encoded semantic unit.

[0098] Illustratively, the encoding process may include, but is not limited to, marking the type and / or content of the semantic unit.

[0099] The type of the semantic unit may include but is not limited to the category to which the semantic unit belongs, for example, "question", "explanation", "motivation" and the like.

[0100] The content of a semantic unit may refer to the specific information content of the semantic unit, such as providing key information, professional knowledge (referring to concepts in a certain field), concept clarification, inspirational content, etc.

[0101] In one example, the conversation analysis module may further include a first generation submodule, which may be used to sort the encoded semantic units in chronological order, generally in chronological order from earliest to latest, so as to form a conversation sequence.

[0102] In one example, the conversation analysis module may further include a matching submodule, which may be used to compare the conversation sequence with a preset conversation pattern in an expert rule library to obtain the conversation analysis result.

[0103] Exemplarily, the preset dialogue modes in the expert rule library may include but are not limited to at least one of the following: cooperation mode; competition mode; conflict mode; emotional support mode; exploration mode; and evaluation mode.

[0104] The cooperation mode may refer to cooperation among multiple learners, including but not limited to at least one of the following:

[0105] Common goals: Learners have common learning goals and actively participate in discussions, support each other, and complete tasks together;

[0106] Active interaction: learners actively interact with each other, such as asking questions, explaining, clarifying, giving feedback, etc.

[0107] Resource sharing: learners share knowledge and resources, such as providing information, explaining concepts, sharing experiences, etc.

[0108] Solve problems through negotiation: Learners solve problems encountered in the learning process through negotiation and discussion.

[0109] The competition mode may refer to a competition relationship between multiple learners, including but not limited to at least one of the following:

[0110] Personal goals: Learners are more concerned with achieving personal goals, such as competing for points, gaining recognition, etc.

[0111] Lack of cooperation: Lack of cooperation among learners, such as competing with each other, hiding information from each other, etc.

[0112] Negative interactions: Learners interact negatively with each other, such as ridicule, belittling, and refusal to cooperate.

[0113] The conflict mode can refer to disagreements among multiple learners, including but not limited to at least one of the following:

[0114] Disagreement: Learners differ in their opinions and cannot reach a consensus.

[0115] Emotional agitation: Learners experience emotional agitation among themselves, such as quarreling, accusations, and attacks.

[0116] Lack of Communication: Learners lack effective communication and are unable to understand each other’s perspectives.

[0117] The emotional support model refers to the emotional expression of support and encouragement between multiple learners, including but not limited to at least one of the following:

[0118] Encouragement and support: Learners encourage and support each other, such as praise, affirmation, comfort, etc.

[0119] Expressing emotions: Learners express their emotions to each other, such as sharing happiness, expressing concerns, etc.

[0120] Building Trust: Build trust among learners, such as listening, understanding, respect, etc.

[0121] The exploration mode refers to multiple learners exploring or conducting in-depth research on a problem, including but not limited to at least one of the following:

[0122] Propose hypotheses: Learners propose hypotheses among themselves and conduct exploration and verification.

[0123] Ask questions: Learners ask questions to each other and find answers.

[0124] Divergent thinking: Learners engage in divergent thinking and put forward different viewpoints and ideas.

[0125] The assessment mode may refer to various assessments conducted among multiple learners, including but not limited to at least one of the following:

[0126] Self-assessment: Learners assess themselves and reflect on their learning process and outcomes.

[0127] Peer Assessment: Learners conduct peer assessment among themselves to evaluate each other’s learning process and learning outcomes.

[0128] Teacher Assessment: Teachers assess learners and provide feedback and guidance.

[0129] Exemplarily, the conversation analysis result may include at least one of the following: feature information of the conversation sequence; and problems existing in the conversation.

[0130] The characteristic information of the dialogue sequence may include but is not limited to at least one of the following: turn structure; interactive relationship between learners; sentiment analysis, etc.

[0131] Among them, turn structure can refer to analyzing the composition and order of different turns in a conversation, such as question-answer, explanation-clarification, suggestion-feedback, etc.

[0132] Among them, the interactive relationships between learners can be, for example, leading-following, supporting-opposing, collaborating-competing, etc.

[0133] Among them, sentiment analysis can be used to analyze the learner's emotional expression in the conversation, such as positive, negative, neutral, etc.

[0134] Among them, problems in the conversation may refer to certain potential learning problems or bad interaction patterns, including but not limited to at least one of the following: repetitive questions; unresolved differences of opinion.

[0135] The above is only an exemplary description, and the present disclosure does not limit the specific content of the conversation analysis results.

[0136] In some embodiments, the feedback module 103 can be used to automatically generate feedback information through a large language model based on the dialogue analysis results. The feedback information can be used to feedback problems in the dialogue. The problems in the dialogue can refer to some potential learning problems or bad interaction patterns, including but not limited to at least one of the following: repetitive questions; unresolved differences of opinion.

[0137] In one example, the feedback module 103 may include a problem detection submodule, which may be used to detect and mark problems existing in the above conversation based on the above conversation analysis results.

[0138] In one example, the feedback module 103 may include a second generating submodule, which may be used to generate feedback information based on the problem existing in the conversation.

[0139] Exemplarily, the second generating submodule may also be used to generate guiding questions and / or prompt information.

[0140] Exemplarily, guiding questions are used to guide the multiple learners to solve the problems existing in the above dialogue.

[0141] For example, when a learner asks repeated questions, the second generation submodule can generate guiding questions such as "Try to summarize the previous discussion to see if some key information is missing" and "Please try to express your views on this issue."

[0142] For example, you can generate guiding questions based on the goals of collaborative learning, such as: "Do other students have different opinions?", "Do you think there is room for improvement in this method?", etc.

[0143] For example, when a discussion reaches a deadlock or conflict occurs, you can generate guiding questions to ease the conflict, such as: "Can we look at this issue from a different angle?" or "Listen to everyone's opinions before making a decision."

[0144] Exemplarily, the prompt information may be used to prompt the multiple learners about the problems existing in the conversation.

[0145] In the above embodiment, the dialogues between multiple learners in the collaborative learning process can be analyzed in real time, and problems in the dialogues can be fed back. The dynamic analysis is highly real-time, which enhances the recognition and adaptability of dynamic interactions and improves the usability for diverse learning situations.

[0146] In some embodiments, Figure 2 is based on Figure 1 The structural diagram of another conversation analysis system provided by the embodiment shown is as follows. Figure 2 As shown, the system may also include:

[0147] Data collection module 104.

[0148] In one example, the data collection module 104 can be used to acquire multi-modal data.

[0149] Among them, multimodal data can refer to data in multiple formats, including but not limited to audio format, video format, and text format.

[0150] In one example, the data collection module 104 may collect audio data and / or video data of the learners in real time through audio and video collection equipment, so as to capture their voice communication, facial expressions and interactive situations.

[0151] Exemplarily, the audio and video acquisition equipment includes but is not limited to a microphone, a camera, etc.

[0152] In one example, the data collection module 104 may obtain text data input by the learner through a text input interface, including but not limited to online discussion records, system operation logs, etc.

[0153] In the above embodiment, the system can synchronously collect data in multiple formats in real time, ensure the integrity of the data, and improve the real-time performance of data collection.

[0154] In some embodiments, Figure 3 is based on Figure 1 The structural diagram of another conversation analysis system provided by the embodiment shown is as follows. Figure 3 As shown, the system may also include:

[0155] Evaluation module 105.

[0156] In one example, the evaluation module 105 may be used to evaluate the interaction between the multiple learners, and generate a learning evaluation report based on the conversation analysis result and the evaluation result.

[0157] Exemplarily, the evaluation module 105 may include an evaluation submodule, which may evaluate the conversation content of each learner based on preset evaluation indicators and determine an evaluation result, which may include a score for evaluating parameters such as the interaction quality and participation enthusiasm of each learner.

[0158] Exemplarily, the evaluation module 105 may include a report generation submodule, which may be used to automatically summarize the system's conversation analysis results and the above-mentioned evaluation results to generate a learning evaluation report to help learners and educators understand the performance and shortcomings of each learner in the collaborative learning process.

[0159] In the above embodiment, a detailed evaluation report can be provided, the cooperation experience between learners can be optimized, more efficient knowledge construction can be promoted, and more efficient learning goals can be achieved, and the usability is high.

[0160] Figure 4 FIG. 1 is a flow chart of a conversation analysis method provided by an exemplary embodiment. Figure 4 As shown, the method may include the following steps:

[0161] Step S401 , semantic segmentation is performed on the data collected in real time to obtain a plurality of semantic units.

[0162] In some embodiments, the process of obtaining multiple semantic units is similar to the action performed by the data processing module 101 and will not be repeated here.

[0163] Step S402: after a dialogue sequence is formed based on the multiple semantic units, the dialogue sequence is compared with a preset dialogue pattern in an expert rule library to obtain a dialogue analysis result.

[0164] In some embodiments, the process of obtaining the conversation analysis results is similar to the actions performed by the conversation analysis module 102 and will not be repeated here.

[0165] Step S403: generating feedback information based on the conversation analysis result.

[0166] In some embodiments, the process of generating feedback information is similar to the action performed by the feedback module 103 and will not be described in detail here.

[0167] In the above embodiment, the dialogues between multiple learners in the collaborative learning process can be analyzed in real time, and problems in the dialogues can be fed back. The dynamic analysis is highly real-time, which enhances the recognition and adaptability of dynamic interactions and improves the usability for diverse learning situations.

[0168] Figure 5 is based on Figure 4The embodiment shown is a flow chart of another method for conversation analysis. Figure 5 As shown, before executing step S401, the method may further include the following steps:

[0169] Step S501, performing preprocessing on the data.

[0170] In some embodiments, the preprocessing process is similar to the actions performed by the preprocessing submodule and will not be described in detail here.

[0171] In the above embodiment, the text sequence can be preprocessed to improve the reliability of classroom conversation analysis.

[0172] Figure 6 is based on Figure 4 The embodiment shown is a flow chart of another method for conversation analysis. Figure 6 As shown, the method may include the following steps:

[0173] Step S601, collecting the data.

[0174] In some embodiments, the process of collecting data is similar to the actions performed by the data collection module 104 and will not be repeated here.

[0175] In the above embodiment, data in various formats can be collected synchronously and in real time, thereby ensuring the integrity of the data and improving the real-time performance of data collection.

[0176] Figure 7 is based on Figure 4 The embodiment shown is a flow chart of another method for conversation analysis. Figure 7 As shown, the method may include the following steps:

[0177] Step S701, evaluating the interaction between the multiple learners to obtain an evaluation result, and generating a learning evaluation report based on the conversation analysis result and the evaluation result.

[0178] In some embodiments, the process of generating a learning evaluation report is similar to the actions performed by the evaluation module 105 and will not be described in detail here.

[0179] In the above embodiment, a detailed evaluation report can be provided, the cooperation experience between learners can be optimized, more efficient knowledge construction can be promoted, and more efficient learning goals can be achieved, and the usability is high.

[0180] The above process is further illustrated by the following examples.

[0181] In the disclosed embodiment, a collaborative learning dialogue recognition and feedback system based on real-time analysis is provided. The main goal of the system is to achieve intelligent analysis and intervention of learner dialogues in the collaborative learning process through multimodal data collection, real-time dialogue analysis, dynamic feedback and quality assessment. The system is generally divided into five modules, namely data collection, data processing, dialogue analysis, dynamic feedback and quality assessment.

[0182] For example Figure 8 As shown, the system may include the following modules:

[0183] Data acquisition module 801:

[0184] Functional description: This module is mainly responsible for collecting multimodal data from learners' various interactive activities. Data sources include audio, video, text input and system operation records, etc., to fully cover all kinds of communication information in the learning process.

[0185] Audio and video acquisition equipment: Use microphones, cameras and other equipment to collect learners’ audio and video data in real time to capture their voice communication, facial expressions and interactive situations.

[0186] Text input interface: used to obtain learners' text input data, such as online discussion records, system operation logs, etc.

[0187] Specifically, the multimodal data acquisition technology supported by this module can synchronously collect data in multiple formats to ensure the comprehensiveness and real-time nature of the data.

[0188] Data processing module 802:

[0189] Function description: Standardize the collected data to provide a high-quality data source for subsequent analysis.

[0190] Data preprocessing submodule: This submodule performs noise reduction on audio data and segments video content to extract relevant features. It also cleans and standardizes text data to remove redundant information.

[0191] Dialogue segmentation submodule: Using natural language processing technology, the learner's continuous dialogue is segmented into several semantically independent expression units, providing basic units for subsequent dialogue analysis.

[0192] Specifically, the data processing module can combine natural language processing technology and signal processing algorithms to ensure the accuracy of data segmentation and the efficiency of processing.

[0193] Conversation analysis module 803:

[0194] Functional description: Through intelligent algorithms and large language models, the system recognizes and analyzes learners' conversation patterns and generates conversation sequences.

[0195] Dialogue coding submodule: Based on the inductive collaborative learning dialogue coding framework, each expression unit is coded, marking its category (such as "question", "explanation", "motivation", etc.), and whether it contains professional content.

[0196] Sequential pattern matching submodule: Organizes the encoded expression units into a dialogue sequence in chronological order, and compares them with the preset patterns in the expert rule base to analyze the characteristics and potential problems of the current dialogue sequence.

[0197] Specifically, the dialogue analysis module can combine the collaborative learning dialogue coding framework generated based on the literature review with the collaborative learning dialogue sequence expert rule base to realize the automatic recognition of dialogue patterns through intelligent algorithms.

[0198] Dynamic feedback module 804:

[0199] Functional description: Based on the results of conversation analysis, the system can dynamically generate feedback and guidance information.

[0200] Real-time problem detection submodule: Automatically detects problems in the current conversation sequence, such as repeated questions, unresolved differences of opinion, etc., and marks these problems as requiring intervention.

[0201] Guiding conversation prompt submodule: Based on the identified problems, the system generates guiding questions and prompt information to assist learners in conducting more in-depth discussions and optimize the efficiency of the discussions.

[0202] Specifically, the dynamic feedback module can rely on a large language model (such as GPT) to provide learners with personalized and real-time guidance through natural language understanding and dialogue generation technology.

[0203] Quality Assessment Module 805:

[0204] Function description: The system automatically evaluates the quality of learners' interactions and generates personalized learning reports.

[0205] Learner dialogue evaluation submodule: Based on the preset evaluation indicators, the dialogue content of each learner is scored to evaluate their interaction quality, participation enthusiasm, etc.

[0206] Learning report generation submodule: automatically summarizes the results of system analysis and evaluation, generates a learning report for each learner, and helps learners and educators understand the performance and deficiencies in collaborative learning.

[0207] Specifically, the quality assessment module can provide comprehensive quality assessment and detailed feedback reports based on data analysis algorithms and learning evaluation models.

[0208] In the above embodiment, the system comprehensively covers the data collection, analysis, feedback and evaluation process in cooperative learning through the mutual cooperation of the above five modules. Through the combination of multimodal data collection and natural language processing technology, the system realizes accurate recognition and real-time feedback of dialogues in cooperative learning, optimizes the learners' cooperative experience, and promotes more efficient knowledge construction and achievement of learning goals.

[0209] The disclosed embodiments significantly improve the recognition efficiency and accuracy of dialogue sequences in the collaborative learning process. Compared with the existing retrospective analysis method, it can provide real-time feedback during the learner interaction process, helping learners to adjust the content of the dialogue in a timely manner, thereby improving the overall efficiency and effectiveness of collaborative learning. By integrating machine learning and natural language processing technology, the system breaks through the limitations of static coding, can dynamically capture diverse interaction modes, and significantly enhances the system's adaptability to complex learning situations.

[0210] In addition, the disclosed embodiment combines a variety of educational theories to design a multi-level, scalable coding framework, which enables the system to provide more educationally valuable feedback and guidance, further improving learners' collaborative ability. By reducing the reliance on a single data set and introducing adaptive learning strategies, the adaptability and scalability of the system have also been significantly improved, allowing it to be flexibly applied in different cultural backgrounds and task objectives.

[0211] In summary, the disclosed embodiments can achieve a comprehensive improvement in the effectiveness of collaborative learning by improving real-time performance, dynamic recognition capabilities, situational adaptability, and deep integration of educational theories, and address the shortcomings of existing technologies in real-time feedback, dialogue content recognition, and adaptability to diverse scenarios.

[0212] Regarding the technical solution in 2, are there other alternative solutions that can also achieve the purpose of the invention?

[0213] In some embodiments, the following solutions may also be used for conversation analysis and intervention:

[0214] Solution 1, static dialogue encoding and classification system.

[0215] Such systems usually use offline methods to label and classify students’ conversations based on recorded videos or conversation data. Although this method can provide basic analysis and summary for collaborative learning, its disadvantage is that it lacks real-time performance and cannot capture and feedback students’ interactive changes in real time. Therefore, this static coding method can only provide limited improvement suggestions through retrospective analysis after the learning activity is over.

[0216] Solution 3: Data monitoring and recording functions in learning management system.

[0217] Some learning management systems (LMS) provide monitoring and recording functions for student conversations, such as tracking and archiving postings, comments, and online collaborative document editing. These systems can record the quantity and time of student conversations, thereby analyzing some of the conversation content of collaborative learning. However, such systems mainly focus on data collection and archiving, but lack intelligent analysis and in-depth understanding. LMS often cannot effectively identify the quality of interaction or the sequence characteristics of conversations, nor can they provide the ability to intervene in real time.

[0218] Although the above scheme can achieve the goal of collaborative learning dialogue analysis to some extent, it still has shortcomings in real-time feedback, dynamic analysis, intelligent recognition and interactive intervention.

[0219] In the disclosed embodiments, a real-time dialogue recognition and feedback system is provided, which dynamically recognizes learning dialogues through intelligent algorithms and provides instant feedback to learners to help them optimize the cooperation process and improve learning outcomes.

[0220] In addition, in the embodiments of the present disclosure, a dynamic interaction recognition algorithm is adopted, for example, a dynamic interaction recognition algorithm based on machine learning and natural language processing technology is introduced, which can flexibly respond to complex and changeable learning situations. It can break through the limitations of traditional static labels and recognize complex dialogue sequences in different situations through deep learning models.

[0221] In addition, in the disclosed embodiment, a multi-level and scalable dialogue coding framework is adopted: by integrating multiple educational theories, the present invention constructs a coding framework that includes multiple dimensions such as cognitive regulation, social interaction, and emotional support, so that the system has higher adaptability and scalability in different learning situations.

[0222] Example 1, scenario: group discussion class.

[0223] Learners: Xiaoming, Xiaohong, Xiaohua and Xiaoli.

[0224] Learning objectives: Explore methods of biodiversity conservation.

[0225] System functions:

[0226] Data collection: Collect learners' voice, video and text data through microphone, camera and text input interface.

[0227] Data processing: pre-process and segment the collected data to extract semantically independent expression units.

[0228] Conversation analysis: Conversation encoding and sequence pattern matching to identify learners’ interaction patterns and conversational features.

[0229] Dynamic feedback: Real-time question detection and guided conversation prompts help learners optimize conversation content.

[0230] Quality Assessment: Evaluate the quality of learners’ interactions and generate personalized learning reports.

[0231] Implementation process:

[0232] Data collection: The system collects learners' voice, video and text data in real time.

[0233] Data processing: The system pre-processes and segments the collected data, dividing the learner's continuous dialogue into several semantically independent expression units.

[0234] Conversation Analysis:

[0235] Dialogue coding: The system codes each expression unit, marking its category (such as questioning, explanation, negotiation, motivation, etc.) and whether it contains professional content.

[0236] Sequence pattern matching: The system organizes the encoded expression units into dialogue sequences in chronological order, and compares them with the preset patterns in the expert rule base to analyze the characteristics and potential problems of the current dialogue sequence.

[0237] Dynamic Feedback:

[0238] Real-time problem detection: The system detected that Xiaohong and Xiaohua had different opinions but failed to reach a consensus through effective communication.

[0239] Guiding dialogue prompt: The system generates a guiding dialogue prompt: "Xiaohong and Xiaohua have different opinions. You can try to put yourself in each other's shoes, understand each other's opinions, and try to find common ground."

[0240] Quality Assessment:

[0241] Learner dialogue evaluation: The system evaluates learners' interaction participation, interaction depth, interaction effect and interaction style.

[0242] Learning report generation: The system generates a learning report for each learner, showing their scores and rankings on different evaluation indicators and providing personalized improvement suggestions.

[0243] Implementation effect:

[0244] Optimized the conversation content: Through guided conversation prompts, Xiao Ming and Xiao Hong finally reached a consensus and jointly proposed a plan to protect biodiversity.

[0245] Improved the effectiveness of collaborative learning: learners actively participated in dialogue, explored issues in depth, and ultimately achieved learning goals.

[0246] Improved learners' collaborative learning ability: Learners discovered their shortcomings and developed improvement plans by reflecting on their own interaction patterns and learning reports.

[0247] The above embodiments demonstrate the effect of the cooperative learning dialogue recognition and feedback system in practical applications. By real-time monitoring, analysis and intervention of learners' dialogues, the system can help learners optimize the content of dialogues, improve the effect of cooperative learning, and ultimately enhance learners' cooperative learning ability.

[0248] Example 2, scenario: group discussion on an online learning platform.

[0249] Learners: Xiao Li, Xiao Zhang, Xiao Wang, Xiao Zhao.

[0250] Learning Objectives: Discuss the applications of artificial intelligence in healthcare.

[0251] System functions:

[0252] Data collection: Through the text input interface, learners' text input data is collected, such as online discussion records, system operation logs, etc.

[0253] Data processing: Clean and standardize the collected text data to remove redundant information.

[0254] Conversation Analysis:

[0255] Dialogue coding: The system codes each text expression unit, marking its category (such as question, explanation, negotiation, motivation, etc.) and whether it contains professional content.

[0256] Sequence pattern matching: The system organizes the encoded expression units into dialogue sequences in chronological order, and compares them with the preset patterns in the expert rule base to analyze the characteristics and potential problems of the current dialogue sequence.

[0257] Dynamic Feedback:

[0258] Real-time question detection: The system detected that Xiao Li had low participation in the discussion and rarely expressed his own opinions.

[0259] Guiding dialogue prompt: The system generates a guiding dialogue prompt: "Xiao Li, you haven't expressed your opinion yet. You can try to share your views on the application of artificial intelligence in the medical field."

[0260] Quality Assessment:

[0261] Learner dialogue evaluation: The system evaluates learners' interaction participation, interaction depth, interaction effect and interaction style.

[0262] Learning report generation: The system generates a learning report for each learner, showing their scores and rankings on different evaluation indicators and providing personalized improvement suggestions.

[0263] Implementation process:

[0264] Data collection: The system collects text input data from learners on the online learning platform.

[0265] Data processing: The system cleans and standardizes the collected text data.

[0266] Conversation Analysis:

[0267] Dialogue coding: The system codes each text expression unit, marking its category (such as question, explanation, negotiation, motivation, etc.) and whether it contains professional content.

[0268] Sequence pattern matching: The system organizes the encoded expression units into dialogue sequences in chronological order, and compares them with the preset patterns in the expert rule base to analyze the characteristics and potential problems of the current dialogue sequence.

[0269] Dynamic Feedback:

[0270] Real-time question detection: The system detected that Xiao Li had low participation in the discussion and rarely expressed his own opinions.

[0271] Guiding dialogue prompt: The system generates a guiding dialogue prompt: "Xiao Li, you haven't expressed your opinion yet. You can try to share your views on the application of artificial intelligence in the medical field."

[0272] Quality Assessment:

[0273] Learner dialogue evaluation: The system evaluates learners' interaction participation, interaction depth, interaction effect and interaction style.

[0274] Learning report generation: The system generates a learning report for each learner, showing their scores and rankings on different evaluation indicators and providing personalized improvement suggestions.

[0275] Implementation effect:

[0276] Improved learner engagement: Through guided conversation prompts, Xiao Li actively participated in the discussion and shared his own views.

[0277] Enriched discussion content: Learners explored the application of artificial intelligence in the medical field from different perspectives, making the discussion content richer and more in-depth.

[0278] Improved learners' collaborative learning ability: Learners discovered their shortcomings and developed improvement plans by reflecting on their own interaction patterns and learning reports.

[0279] The above embodiments demonstrate the application effect of the collaborative learning dialogue recognition and feedback system on the online learning platform. By real-time monitoring, analysis and intervention of learners' dialogues, the system can help learners improve their participation, enrich discussion content, and ultimately improve learners' collaborative learning ability.

[0280] In some embodiments, the present specification also provides a conversation analysis device, which can be deployed on a server of a conversation analysis system. The device may include: one or more processors; wherein the processor is used to execute the steps of any of the methods described above.

[0281] Fig. 9 This is a schematic diagram of a server of a conversation analysis system provided by an exemplary embodiment. Fig. 9 At the hardware level, the device includes a processor 902, an internal bus 904, a network interface 906, a memory 908, and a non-volatile memory 910, and may also include hardware required for other functions. One or more embodiments of this specification can be implemented based on software. For example, a large model is deployed on the server, and the processor 902 reads the corresponding computer program from the non-volatile memory 910 to the memory 908 and then runs it. Of course, in addition to the software implementation, one or more embodiments of this specification do not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0282] Based on the same concept as the above method, this specification also provides a computer-readable storage medium on which computer instructions are stored. When the instructions are executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0283] Based on the same concept as the above method, this specification also provides a computer program product, including a computer program / instruction, which implements the steps of the method described in any of the above embodiments when executed by a processor.

[0284] In addition, the present disclosure also provides a database operation and maintenance system, which may include a database; a large language model; wherein the model may execute the steps of the method described in any of the above embodiments.

[0285] Those skilled in the art will readily appreciate other embodiments of the specification after considering the specification and practicing the invention claimed herein. The specification is intended to cover any variations, uses or adaptations of the specification that follow the general principles of the specification and include common knowledge or customary techniques in the art that are not claimed in the specification. The specification and examples are to be considered exemplary only, and the true scope and spirit of the specification are indicated by the following claims.

[0286] It should be understood that the present description is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present description is limited only by the appended claims.

[0287] The above description is only a preferred embodiment of this specification and is not intended to limit this specification. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of this specification should be included in the scope of protection of this specification.

Claims

1. A conversation analysis system, characterized in that: The system comprises: A data processing module is used to semantically segment the data collected in real time to obtain multiple semantic units; wherein the data is related to the dialogue between multiple learners in the collaborative learning process; A dialogue analysis module, configured to, after forming a dialogue sequence based on the plurality of semantic units, compare the dialogue sequence with a preset dialogue pattern in an expert rule library to obtain a dialogue analysis result; A feedback module is used to generate feedback information based on the dialogue analysis results; wherein the feedback information is used to feedback problems existing in the dialogue.

2. The system according to claim 1, characterized in that The data processing module comprises: A preprocessing submodule is used to perform preprocessing on the data; wherein the preprocessing includes at least one of the following: Noise reduction processing; Content segmentation processing; A cleaning process is used to remove redundant information.

3. The system according to claim 1, characterized in that The conversation analysis module includes: The dialogue encoding submodule is used to perform encoding processing on each semantic unit to obtain an encoded semantic unit; wherein the encoding processing includes marking the type and / or content of the semantic unit; A first generating submodule is used to sort the encoded semantic units in chronological order to form the dialogue sequence; The matching submodule is used to compare the dialogue sequence with the preset dialogue mode to obtain the dialogue analysis result.

4. The system according to claim 1, characterized in that The feedback module comprises: A problem detection submodule, used for detecting problems existing in the conversation based on the conversation analysis result; The second generating submodule is used to generate the feedback information based on the problem existing in the dialogue.

5. The system according to claim 4, characterized in that The second generation submodule is also used for: Based on the problem existing in the conversation, a guiding question and / or prompt information is generated; wherein the guiding question is used to guide the multiple learners to solve the problem, and the prompt information is used to prompt the multiple learners about the problem existing in the conversation.

6. The system according to claim 1, characterized in that The system further comprises: A data collection module is used to collect the data; wherein the format of the data is at least one of the following: Audio format; Video format; Text format.

7. The system according to any one of claims 1 to 6, characterized in that: The system further comprises: The evaluation module is used to evaluate the interaction between the multiple learners, obtain evaluation results, and generate a learning evaluation report based on the conversation analysis results and the evaluation results.

8. A conversation analysis method, characterized in that: The method comprises: Semantically segmenting the data collected in real time to obtain a plurality of semantic units; wherein the data is related to the dialogue between a plurality of learners in the collaborative learning process; After a dialogue sequence is formed based on the multiple semantic units, the dialogue sequence is compared with a preset dialogue pattern in an expert rule library to obtain a dialogue analysis result; Based on the dialogue analysis result, feedback information is generated; wherein the feedback information is used to feedback problems existing in the dialogue.

9. The method according to claim 8, characterized in that After the dialog sequence is formed based on the multiple semantic units, the dialog sequence is compared with a preset dialog mode in an expert rule library to obtain a dialog analysis result, including: Performing encoding processing on each semantic unit to obtain an encoded semantic unit; wherein the encoding processing includes marking the type and / or content of the semantic unit; Sorting the encoded semantic units in chronological order to form the dialogue sequence; The dialogue sequence is compared with the preset dialogue mode to obtain the dialogue analysis result.

10. The method according to claim 8 or 9, characterized in that: The method further comprises: Based on the problem existing in the conversation, a guiding question and / or prompt information is generated; wherein the guiding question is used to guide the multiple learners to solve the problem, and the prompt information is used to prompt the multiple learners about the problem existing in the conversation.

Images