Method and device for automatic analysis of dialogue flow

Abstract

The present invention discloses a method and device for automatically analyzing a conversation flow, the method comprising: obtaining the conversation data flow of each dialogue party in a conversation to be detected; inputting the conversation data flow of each dialogue party into an analysis engine corresponding to the identity of the dialogue party, the analysis engine identifying key elements of the conversation data flow, and determining the label information of the dialogue party based on the analysis model in the analysis engine; determining the conversation analysis result of the dialogue party according to the label information of the dialogue party. By using the present invention, automatic analysis of the conversation can be realized, which greatly saves manpower and time costs; and the quality of the analysis result can be guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of information processing, and in particular to a method and device for automatically analyzing a dialog flow. Background Art

[0002] With the development of interpersonal interaction tools and technologies, members of society can not only communicate face-to-face, but also remotely through text, voice, video and other means. In some companies, relevant personnel of the company need to insert the services provided by the company into the process of dialogue and communication with the other party. For example, the marketing personnel of the company need to recommend advertising and marketing products to customers every day. Accordingly, the company needs to check the service content of relevant business personnel from the perspectives of quality control, risk control, and value mining, so as to evaluate their business capabilities, service levels, etc., and to find problems and help business personnel improve their relevant service skills.

[0003] Since quality inspection is limited by the particularity of voice conversations, it currently relies mainly on manual sampling and a small amount of semantic recognition, which wastes manpower costs. Summary of the invention

[0004] The embodiment of the present invention provides a method and device for automatically analyzing a dialog flow to solve the problems existing in the prior art.

[0005] To this end, the present invention provides the following technical solutions:

[0006] A method for automatically analyzing a dialog flow, the method comprising:

[0007] Obtaining the conversation data stream of each party in the conversation to be detected;

[0008] Inputting the conversation data stream of each dialogue party into the analysis engine corresponding to the identity of the dialogue party, the analysis engine identifying key elements of the conversation data stream, and determining the label information of the dialogue party based on the analysis model in the analysis engine;

[0009] The conversation analysis result of the conversation party is determined according to the tag information of the conversation party.

[0010] Optionally, the dialogue parties include: a party to be inspected and a party not to be inspected; and the method further includes:

[0011] According to the set quality inspection rules, the conversation analysis results of the party to be quality inspected and the conversation analysis results of the party not to be quality inspected are respectively integrated to obtain a fusion result;

[0012] The quality inspection evaluation result of the party to be inspected is determined according to the fusion result.

[0013] Optionally, the method further comprises:

[0014] The risky conversations in the conversation data stream of the party to be quality inspected are determined according to the conversation analysis results and the quality inspection evaluation results of the party to be quality inspected, and the risky conversations are marked.

[0015] Optionally, the method further comprises:

[0016] Recording the conversation data flow of the party to be inspected in each conversation, as well as the conversation analysis results and quality inspection evaluation results of the party to be inspected obtained based on the conversation;

[0017] Generate a quality inspection year-on-year view of the party to be inspected based on the recorded conversation analysis results and quality inspection evaluation results of the party to be inspected within a certain period; and / or

[0018] A quality inspection work view is generated based on the conversation analysis results and quality inspection evaluation results of each party to be inspected obtained based on the conversations between different parties to be inspected and non-quality inspection parties with the same identity within a certain period of record.

[0019] Optionally, the quality inspection month-on-month view and the quality inspection worker view are in any of the following forms: a line graph, a bar graph, a pie chart.

[0020] Optionally, the method further comprises: establishing an analysis engine corresponding to the identity of the interlocutor in the following manner:

[0021] Set multiple dialogue scenarios based on the identities of the interlocutors and their corresponding analysis requirements;

[0022] Determine key elements of each dialogue scene and multiple different categories of tags hit by the dialogue scene;

[0023] An analysis model corresponding to different labels is established according to the key elements and their logical relationships.

[0024] Optionally, the party to be detected is a service party; the non-detected party is a client party;

[0025] The analysis requirements corresponding to the service provider include service standards;

[0026] The conversation scenarios corresponding to the service provider include any one or more of the following: failure to promote business, insulting customers, failure to expand contact information, and invalid conversations; the key elements of the conversation scenarios corresponding to the service provider include any one or more of the following: key terms, banned words, selling competing products, and illegal keywords;

[0027] The dialogue scene corresponding to the customer includes any one or more of the following: accepting service, vague hesitation, and refusing service; the key elements of the dialogue scene corresponding to the customer include any one or more of the following: speaking speed, hanging up action, and tone.

[0028] Optionally, the logical relationship includes any one or more of the following: a sequential relationship, an inclusion relationship, a hit relationship, a union relationship, and a selection relationship.

[0029] Optionally, the dialogue is a dialogue between people, or a dialogue between people and machines.

[0030] Optionally, the dialogue is a two-party dialogue, or a multi-party dialogue.

[0031] A device for automatically analyzing a dialog flow, the device comprising:

[0032] A data stream acquisition module, used to acquire the conversation data stream of each dialogue party in the conversation to be detected, and input the conversation data stream of each dialogue party into the analysis engine corresponding to the identity of the dialogue party; the analysis engine identifies key elements of the conversation data stream, and determines and outputs the label information of the dialogue party based on each analysis model in the analysis engine;

[0033] The analysis result acquisition module is used to determine the conversation analysis result of the conversation party according to the label information of the conversation party output by the analysis engine.

[0034] Optionally, the dialogue parties include: a party to be inspected and a party not to be inspected; and the device further includes:

[0035] Rule setting module, used to set quality inspection rules;

[0036] A fusion module, used to fuse the conversation analysis results of the party to be quality inspected and the conversation analysis results of the party not to be quality inspected according to the quality inspection rules to obtain a fusion result;

[0037] The quality inspection and evaluation module is used to determine the quality inspection and evaluation result of the party to be inspected according to the fusion result.

[0038] Optionally, the device further comprises:

[0039] The marking module is used to determine the risky dialogues in the conversation data stream of the party to be quality inspected according to the dialogue analysis results and quality inspection evaluation results of the party to be quality inspected, and mark the risky dialogues.

[0040] Optionally, the device further comprises:

[0041] A recording module, used to record the conversation data flow of the party to be inspected in each conversation, as well as the conversation analysis results and quality inspection evaluation results of the party to be inspected obtained based on the conversation;

[0042] A first view generation module is used to generate a quality inspection year-on-year view of the party to be inspected according to the conversation analysis results and quality inspection evaluation results of the party to be inspected within a certain period recorded by the recording module; and / or

[0043] The second view generation module is used to generate a quality inspection work view based on the conversation analysis results and quality inspection evaluation results of each party to be inspected obtained based on the conversations between different parties to be inspected and non-quality inspection parties with the same identity within a certain period of time.

[0044] Optionally, the quality inspection month-on-month view and the quality inspection worker view are in any of the following forms: a line graph, a bar graph, a pie chart.

[0045] Optionally, the device further comprises an analysis engine establishment module, which is used to establish an analysis engine corresponding to the identity of the interlocutor; the analysis engine establishment module comprises:

[0046] A dialogue scene setting unit, used to set multiple dialogue scenes according to the identities of the dialogue parties and their corresponding analysis requirements;

[0047] An information determination unit, used to determine key elements of each dialogue scene and multiple different categories of tags hit by the dialogue scene;

[0048] The analysis model building unit is used to build analysis models corresponding to different tags according to the key elements and their logical relationships.

[0049] Optionally, the party to be detected is a service party; the non-detected party is a client party;

[0050] The analysis requirements corresponding to the service provider include service standards;

[0051] The conversation scenarios corresponding to the service provider include any one or more of the following: failure to promote business, insulting customers, failure to expand contact information, and invalid conversations; the key elements of the conversation scenarios corresponding to the service provider include any one or more of the following: key terms, banned words, and selling competing products;

[0052] The dialogue scene corresponding to the customer includes any one or more of the following: accepting service, vague hesitation, and refusing service; the key elements of the dialogue scene corresponding to the customer include any one or more of the following: speaking speed, hanging up action, and tone.

[0053] Optionally, the logical relationship includes any one or more of the following: a sequential relationship, an inclusion relationship, a hit relationship, a union relationship, and a selection relationship.

[0054] Optionally, the dialogue is a dialogue between people, or a dialogue between people and machines.

[0055] Optionally, the dialogue is a two-party dialogue, or a multi-party dialogue.

[0056] A computer device, comprising: one or more processors, a memory;

[0057] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the aforementioned method.

[0058] A readable storage medium stores instructions thereon, wherein the instructions are executed to implement the above-mentioned method.

[0059] The method and device for automatically analyzing the conversation flow provided by the embodiment of the present invention, when it is necessary to analyze the conversation, the conversations of the parties therein are divided, that is, the conversation data streams of each conversation party in the conversation to be detected are obtained; the conversation data streams of each conversation party are respectively input into the analysis engine corresponding to the identity of the conversation party, and the analysis engine identifies the key elements of the conversation data stream, and determines the label information of the conversation party based on each analysis model in the analysis engine; the conversation analysis result of the conversation party is determined according to the label information of the conversation party, thereby realizing automatic analysis of the conversation content, greatly saving manpower and time costs; because the corresponding analysis engine is established in combination with the identity of the conversation party, and the data stream of the entire conversation is analyzed, the accuracy and comprehensiveness of the analysis result are effectively improved, and the quality of the analysis result is guaranteed. In addition, the scheme of the present invention can meet the needs of users whether it is for offline conversation analysis or real-time online conversation analysis.

[0060] Furthermore, the automatic conversation flow analysis method and device provided in the embodiments of the present invention can also fuse the conversation analysis results of the party to be quality inspected and the party not to be quality inspected in the conversation according to the set quality inspection rules for application scenarios with quality inspection requirements, and perform corresponding quality inspection evaluation on the party to be quality inspected based on the fusion results to obtain the quality inspection evaluation results.

[0061] Furthermore, the automatic conversation flow analysis method and device provided in the embodiments of the present invention can also determine the risky conversations in the conversation data flow of the party to be quality inspected based on the conversation analysis results and quality inspection evaluation results of the party to be quality inspected, and mark them, thereby reminding the party to be quality inspected to avoid the same problems in subsequent similar conversation scenarios.

[0062] Furthermore, the method and device for automatic analysis of conversation flows provided in the embodiments of the present invention can also record the conversation data flow of the party to be quality inspected in each conversation, as well as the conversation analysis results and quality inspection evaluation results of the party to be quality inspected based on the conversation, and then generate views of multiple different dimensions based on the recorded historical information, more intuitively display the data information related to quality inspection, and provide users with visual data support. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0064] Figure 1 is a flow chart of establishing an analysis engine in an embodiment of the present invention;

[0065] Figure 2 This is a flow chart of a method for automatically analyzing a dialog flow according to an embodiment of the present invention;

[0066] Figure 3 This is a specific application flow chart of the method for automatically analyzing a dialog flow according to an embodiment of the present invention;

[0067] Figure 4 This is a structural block diagram of a device for automatically analyzing a dialog flow according to an embodiment of the present invention;

[0068] Figure 5 is a structural block diagram of an analysis engine establishment module in an embodiment of the present invention;

[0069] Figure 6 This is another structural block diagram of the automatic dialog flow analysis device according to an embodiment of the present invention;

[0070] Figure 7 is a block diagram of a device for a method for automatically analyzing a dialog flow according to an exemplary embodiment;

[0071] Figure 8 It is a schematic diagram of the structure of the server in the embodiment of the present invention. DETAILED DESCRIPTION

[0072] In order to enable persons skilled in the art to better understand the solutions of the embodiments of the present invention, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and implementation modes.

[0073] An embodiment of the invention provides a method and device for automatic analysis of conversation flows. When a conversation needs to be analyzed, the conversations of the parties therein are diverted, that is, the conversation data streams of each party in the conversation to be detected are obtained; the conversation data streams of each party are respectively input into an analysis engine corresponding to the identity of the party, and the analysis engine identifies key elements of the conversation data stream and determines the label information of the party based on each analysis model in the analysis engine; the conversation analysis result of the party is determined according to the label information of the party, thereby realizing automatic analysis of the conversation.

[0074] The solution of the present invention can be applicable to application scenarios of bilateral or multi-party dialogue, and the dialogue can be a human-to-human dialogue or a human-to-machine dialogue.

[0075] In addition, the solution of the present invention is mainly aimed at automatically analyzing the conversations in which one of the dialogue parties has a professional identity (human side) or professional performance (machine side) in a certain scenario and with a certain purpose.

[0076] Conversation analysis with natural persons can help us understand the occupation-related assessment information of the interlocutor, such as business ability, work attitude, service awareness, etc.; conversation analysis with machines can help evaluate whether the dialogue plan designed for the machine's functions has deficiencies or defects, so as to make improvements later and achieve better functional results.

[0077] As mentioned above, the present invention can be applied to application scenarios of two-party dialogue or multi-party dialogue. Regardless of whether it is a two-party dialogue or a multi-party dialogue, usually only one of the parties has a business detection requirement. For the convenience of description, the dialogue party will be referred to as the party to be detected, and the party or parties having a dialogue with the party to be detected will be referred to as non-detection parties.

[0078] Since the conversation reflects the interaction process between two parties, not only can the conversation data stream of the party to be detected be analyzed to obtain effective information reflecting its behavior, ability, etc., but also the conversation data stream of the non-detected party in the conversation can be analyzed to obtain auxiliary information reflecting its behavior, ability, etc. Accordingly, in practical applications, it is necessary to establish a corresponding analysis engine for each interlocutor in the conversation.

[0079] First, the process of establishing an analysis engine in an embodiment of the present invention is described in detail below.

[0080] like Figure 1 FIG. 1 is a flowchart of establishing an analysis engine in an embodiment of the present invention, comprising the following steps:

[0081] Step 101, multiple dialogue scenarios are set according to the identities of the dialogue parties and their corresponding analysis requirements.

[0082] The interlocutor identity refers to the identity given by the interlocutor's occupation, which can also be called professional identity. For example, when the interlocutor is a natural person, his professional identity can be: business promoter, product salesperson, company front desk, customer service, etc. When the interlocutor is a machine, his professional identity can be: customer service, etc.

[0083] Due to the differences in dimensions and contents involved in business testing for different professional identities, the corresponding analysis requirements will also be different. Moreover, even for the same professional identity, the purpose of business testing in different institutions may be different, and the corresponding analysis requirements will also be different.

[0084] It should be noted that the dialogue scenario does not refer to the application scenario of the dialogue, but refers to a dialogue scenario with certain characteristics set according to the identity of the dialoguer and the corresponding analysis requirements. For example, the dialogue scenarios set according to the analysis requirements of the work attitude and promotion ability of the business promoter include: insulting customers, invalid dialogue, etc.

[0085] Step 102, determining the key elements of each dialogue scene and multiple different categories of tags hit by the dialogue scene.

[0086] Specifically, the corresponding vocabulary, action and other information can be collected according to the application field, scenario, etc., and the labels can be classified by elements to determine the labels that hit different dialogue scenarios.

[0087] Taking a two-party conversation as an example, it is assumed that the party to be detected is the service party, and the non-detection party is the client party.

[0088] The analysis requirements corresponding to the service provider may include service standards, etc.; the conversation scenarios corresponding to the service provider may include but are not limited to any one or more of the following: failure to promote business, insulting customers, failure to expand contact information, invalid conversations, etc. The key elements of the conversation scenarios corresponding to the service provider may include but are not limited to any one or more of the following: key terms, banned words, competing product sales, etc. The tags hit by the conversation scenarios include but are not limited to any one or more of the following: peers, competing products, account opening, visits, etc.

[0089] Accordingly, the analysis requirements corresponding to the client may be omitted; the conversation scenarios corresponding to the client may include but are not limited to any one or more of the following: accepting service, vague hesitation, refusing service, etc.; the key elements of the conversation scenarios corresponding to the client include but are not limited to any one or more of the following: speaking speed, hanging up action, tone, etc. The tags hit by the conversation scenarios include but are not limited to any one or more of the following: agree, refuse, satisfied, dissatisfied, etc.

[0090] Of course, in actual applications, the settings of the above-mentioned dialogue scenes, key elements, tags and other elements can be flexibly determined according to the actual application scenarios and user needs, and the embodiments of the present invention are not limited to this.

[0091] For example, when the dialogue application scenario is in the customer service field, the tag information may include but is not limited to any one or more of the following: scolding, visiting, problem recording, understanding needs, seeking suggestions, etc.;

[0092] For another example, when the dialogue application scenario is in the field of education, the tag information may include but is not limited to any one or more of the following: course requirements, age, hobbies, time arrangements, etc.;

[0093] For another example, when the dialogue application scenario is in the medical field, the tag information may include but is not limited to any one or more of the following: age, gender, medical history, medication, allergies, etc.

[0094] Step 103: Establish analysis models corresponding to different tags according to the key elements and their logical relationships.

[0095] Each analysis model may be an analysis model composed of one or more key elements and logical relationships, wherein the logical relationship may include but is not limited to any one or more of the following: a sequential relationship, an inclusion relationship, a hit relationship, a joint relationship, a selection relationship, and the like.

[0096] It should be noted that in order to further ensure the effectiveness and accuracy of the analysis results, the analysis engine can be customized according to specific application scenarios and user needs to make it more targeted.

[0097] In practical applications, the conversation data streams of each party in the conversation can be split out first, that is, their respective conversation data streams are obtained respectively, and then the conversation data stream of each party is input into the analysis engine corresponding to the party. The analysis engine automatically analyzes the conversation stream and outputs the corresponding analysis results.

[0098] It should be noted that, in practical applications, the analysis engine may include one or more analysis models, each of which corresponds to one or more different tags. The tags may be determined according to different application scenarios, which is not limited in the embodiments of the present invention.

[0099] like Figure 2 FIG. 1 is a flow chart of a method for automatically analyzing a dialog flow according to an embodiment of the present invention, comprising the following steps:

[0100] Step 201: Acquire the conversation data stream of each party in the conversation to be detected.

[0101] It should be noted that, in actual applications, different methods may be used to obtain the conversation data stream of each conversation party according to different application scenarios and data source forms.

[0102] For example, when a conversation is in progress, each party to the conversation is recorded separately through different audio tracks, thereby obtaining the conversation data stream of each party.

[0103] For another example, if the conversation data source is an audio file containing the entire conversation process, some existing related technologies can be used to split it and obtain respective conversation data streams.

[0104] Step 202, the conversation data stream of each dialogue party is input into the analysis engine corresponding to the identity of the dialogue party, and the analysis engine identifies key elements of the conversation data stream and determines the label information of the dialogue party based on the analysis model in the analysis engine.

[0105] Since key elements may be of different types, such as text, audio, and statistical types, the analysis engine needs to perform audio signal analysis, speech recognition, word segmentation, and other processing on the input conversation data stream to obtain the corresponding audio information such as tone and voice, and statistical information such as duration and speaking speed. In addition, through word segmentation, the engine can obtain various keywords, logical relationships, and hit tag words.

[0106] It should be noted that each tag can correspond to multiple specific tag words. To this end, a corresponding tag word library can be established, and the tag word and the tag of the tag word can be determined by searching the tag word library according to the specific words obtained by word segmentation.

[0107] The identified key elements are matched with each analysis model, and the corresponding label information can be obtained according to the successfully matched analysis model.

[0108] Step 203: Determine the conversation analysis result of the conversation party according to the tag information of the conversation party.

[0109] Specifically, the conversation parties may be classified according to their respective label information to generate analysis results; or analysis results may be generated according to set fusion rules, such as weighted calculation of the label information of the conversation parties; of course, other forms of analysis results may also be used, which is not limited in this embodiment of the present invention.

[0110] The method for automatically analyzing the conversation flow provided by the embodiment of the present invention, when it is necessary to analyze the conversation, diverts the conversations of the parties therein, that is, obtains the conversation data streams of each party in the conversation to be detected; inputs the conversation data streams of each party into the analysis engine corresponding to the identity of the party, and the analysis engine identifies the key elements of the conversation data stream, and determines the label information of the party based on the analysis models in the analysis engine; determines the conversation analysis result of the party according to the label information of the party, thereby realizing automatic analysis of the conversation content, greatly saving manpower and time costs; because the corresponding analysis engine is established in combination with the identity of the interlocutor, and the data stream of the entire conversation is analyzed, the accuracy and comprehensiveness of the analysis result are effectively improved, and the quality of the analysis result is guaranteed. In addition, the scheme of the present invention can meet the needs of users whether it is for offline conversation analysis or real-time online conversation analysis.

[0111] In application scenarios where quality inspection is required for one party in a conversation, the conversation analysis results of the party to be inspected and the conversation analysis results of the party not to be inspected can also be considered simultaneously to conduct a comprehensive evaluation of the conversation quality of the party to be inspected to ensure the accuracy of the evaluation results.

[0112] like Figure 3 FIG. 1 is a specific application flow chart of the method for automatically analyzing a dialog flow according to an embodiment of the present invention, and includes the following steps:

[0113] Step 301: Acquire the conversation data stream of the party to be quality inspected and the conversation data stream not to be quality inspected in the conversation.

[0114] Step 302: Input the conversation data stream of the party to be quality inspected and the conversation data stream not to be quality inspected into their respective corresponding analysis engines, and the analysis engines identify key elements of the conversation data streams and determine the label information of the conversation parties based on the analysis model in the analysis engines.

[0115] Step 303: determining the conversation analysis result of the party to be quality inspected according to the label information of the party to be quality inspected, and determining the conversation analysis result of the party not to be quality inspected according to the label information of the party not to be quality inspected.

[0116] Step 304: According to the set quality inspection rules, the conversation analysis results of the party to be quality inspected and the conversation analysis results of the party not to be quality inspected are respectively integrated to obtain a fusion result.

[0117] The specific fusion method is not limited in the embodiment of the present invention. For example, weighted statistics can be performed according to the tag categories to obtain scores for each type of tag.

[0118] Step 305: Determine the quality inspection evaluation result of the party to be inspected according to the fusion result.

[0119] The quality inspection and evaluation results may be expressed in the form of scores, or in the form of grades (such as excellent, good, fair, poor, etc.), or in the form of scores of different label categories, and there is no limitation on this.

[0120] Furthermore, in another embodiment of the method of the present invention, the risk dialogue in the conversation data stream of the party to be inspected can also be determined based on the dialogue analysis results and quality inspection evaluation results of the party to be inspected, and the risk dialogue can be marked. For example, the party to be inspected is a business party, and the generation of customer complaints (which will lead to customer loss or business revenue loss) is defined as a risk. The dialogue analysis results of the party to be inspected include the label "invalid dialogue", and the dialogue analysis results of the non-quality inspection party include the label "marketing aversion". The quality inspection evaluation result of the party to be inspected is "poor", then it is determined that there is a risk dialogue in the conversation data stream of the party to be inspected, and the risk dialogue corresponds to a dialogue in which the customer has the intention to complain or clearly expresses the intention to complain. For example, the dialogue in which the customer has the intention to complain is: I don't need you for the time being, so don't bother me; the dialogue in which the customer clearly expresses the intention to complain is: You have harassed me and I want to complain about you. Marking the corresponding risk dialogue can remind the party to be inspected to avoid the same problem in subsequent similar dialogue scenarios.

[0121] In application scenarios involving work such as quality inspection and assessment, the work of quality inspection and assessment is usually characterized by continuity and individual diversity. Therefore, considering that in such application scenarios, in order to make the quality inspection results more intuitive and facilitate users to have a comprehensive understanding of the overall quality inspection results, in another embodiment of the method of the present invention, the conversation data flow of the party to be inspected in each conversation, as well as the conversation analysis results and quality inspection evaluation results of the party to be inspected obtained based on the conversation can also be recorded; according to the conversation analysis results and quality inspection evaluation results of the party to be inspected within a certain period of the record, a quality inspection year-on-year view of the party to be inspected is generated; and / or according to the conversation analysis results and quality inspection evaluation results of each party to be inspected obtained based on the conversations between different parties to be inspected and non-quality inspected parties of the same identity within a certain period of the record, a quality inspection work view is generated. The quality inspection year-on-year view and the quality inspection worker view are in any of the following forms: a line graph, a bar graph, a pie chart, etc.

[0122] Accordingly, an embodiment of the present invention also provides a device for automatically analyzing a dialog flow, such as Figure 4 , which is a structural block diagram of a dialog flow automatic analysis device according to an embodiment of the present invention.

[0123] In this embodiment, the device includes the following modules:

[0124] The data stream acquisition module 401 is used to acquire the conversation data stream of each dialogue party in the dialogue to be detected, and input the conversation data stream of each dialogue party into the analysis engine corresponding to the identity of the dialogue party;

[0125] The analysis engine 400 is used to identify key elements of the conversation data stream, and determine and output label information of the conversation parties based on each analysis model in the analysis engine;

[0126] The analysis result acquisition module 402 is used to determine the conversation analysis result of the conversation party according to the tag information of the conversation party output by the analysis engine.

[0127] It should be noted that the analysis engine 400 includes one or more analysis models, each of which corresponds to one or more different tags. The analysis engine 400 can be established by a corresponding analysis engine establishment module. The analysis engine establishment module can be a part of the device of the present invention, or it can be independent of the device of the present invention, which is not limited in this embodiment of the present invention.

[0128] A specific structure of the analysis engine establishment module is as follows Figure 5 As shown, it includes the following units:

[0129] A dialogue scene setting unit 411, used to set multiple dialogue scenes according to the identities of the dialogue parties and their corresponding analysis requirements;

[0130] An information determination unit 412, configured to determine key elements of each dialogue scene and multiple different types of tags hit by the dialogue scene;

[0131] The analysis model establishing unit 413 is used to establish analysis models corresponding to different tags according to the key elements and their logical relationships.

[0132] It should be noted that the dialogue scenario refers to a dialogue scenario with certain characteristics set according to the identity of the dialoguer and the corresponding analysis requirements. For example, the dialogue scenarios set according to the analysis requirements of the work attitude and promotion ability of the business promoter include: insulting customers, invalid dialogue, etc.

[0133] Accordingly, the information determination unit 412 can specifically collect corresponding vocabulary, action and other information according to application fields, scenarios and the like, and classify labels by elements to determine labels hit by different dialogue scenarios.

[0134] Of course, in actual applications, the settings of the above-mentioned dialogue scenes, key elements, tags and other elements can be flexibly determined according to the actual application scenarios and user needs, and the embodiments of the present invention are not limited to this.

[0135] In the embodiment of the present invention, each analysis model may be an analysis model composed of one or more key elements and logical relationships, wherein the logical relationship may include but is not limited to any one or more of the following: sequential relationship, inclusion relationship, hit relationship, joint relationship, selection relationship, etc.

[0136] It should be noted that in order to further ensure the effectiveness and accuracy of the analysis results, the analysis engine can be customized according to specific application scenarios and user needs to make it more targeted.

[0137] The automatic conversation flow analysis device provided by the embodiment of the present invention, when it is necessary to analyze a conversation, diverts the conversations of the parties therein, that is, obtains the conversation data stream of each conversation party in the conversation to be detected; inputs the conversation data stream of each conversation party into the analysis engine corresponding to the identity of the conversation party, and the analysis engine identifies key elements of the conversation data stream and determines the label information of the conversation party based on each analysis model in the analysis engine; determines the conversation analysis result of the conversation party according to the label information of the conversation party, thereby realizing automatic analysis of the conversation content, greatly saving manpower and time costs; because the corresponding analysis engine is established in combination with the identity of the conversation party, and the data stream of the entire conversation is analyzed, the accuracy and comprehensiveness of the analysis result are effectively improved, and the quality of the analysis result is guaranteed.

[0138] Furthermore, if Figure 6 As shown, in another embodiment of the device of the present invention, the following modules may also be included:

[0139] A rule setting module 501 is used to set quality inspection rules;

[0140] A fusion module 502 is used to fuse the conversation analysis results of the party to be quality inspected and the conversation analysis results of the party not to be quality inspected according to the quality inspection rule to obtain a fusion result;

[0141] The quality inspection and evaluation module 503 is used to determine the quality inspection and evaluation result of the party to be quality inspected according to the fusion result.

[0142] The automatic dialog flow analysis device provided in the embodiment of the present invention can integrate the dialog analysis results of the party to be inspected and the party not to be inspected in the dialog, and provide a quality inspection evaluation result for application scenarios with quality inspection requirements.

[0143] Furthermore, in another embodiment of the device of the present invention, a marking module (not shown) may be included, which is used to determine the risky dialogues in the conversation data stream of the party to be inspected according to the dialogue analysis results and the quality inspection evaluation results of the party to be inspected, and mark the risky dialogues. By marking the risky dialogues, the party to be inspected can be reminded to avoid the same problems in subsequent similar dialogue scenarios, thereby improving the dialogue effect.

[0144] Furthermore, in another embodiment of the device of the present invention, the following modules may also be included:

[0145] A recording module, used to record the conversation data flow of the party to be inspected in each conversation, as well as the conversation analysis results and quality inspection evaluation results of the party to be inspected obtained based on the conversation;

[0146] A first view generation module is used to generate a quality inspection year-on-year view of the party to be inspected according to the conversation analysis results and quality inspection evaluation results of the party to be inspected within a certain period recorded by the recording module; and / or

[0147] The second view generation module is used to generate a quality inspection work view based on the conversation analysis results and quality inspection evaluation results of each party to be inspected obtained based on the conversations between different parties to be inspected and non-quality inspection parties with the same identity within a certain period of time.

[0148] It should be noted that the quality inspection month-on-month view and the quality inspection worker view may be in, but not limited to, any of the following forms: a line graph, a bar graph, a pie chart, etc. The specific form of the view may be set according to user needs.

[0149] The automatic dialog flow analysis device of this embodiment presents the quality inspection and evaluation results of different dimensions through views, thereby providing visual data support for users.

[0150] The automatic conversation flow analysis device of the embodiment of the present invention can be applied to the application scenario of two-party or multi-party conversation, and the conversation can be a person-to-person conversation or a person-to-computer conversation. Moreover, the automatic conversation flow analysis device provided by the embodiment of the present invention can meet the needs of users regardless of whether it is for offline conversation analysis or real-time online conversation analysis.

[0151] Figure 7 8 is a block diagram of an apparatus 800 for automatically analyzing a dialog flow according to an exemplary embodiment. For example, the apparatus 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0152] Reference Figure 7 , the device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / output (I / O) interface 812 , a sensor component 814 , and a communication component 816 .

[0153] The processing component 802 generally controls the overall operation of the device 800, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

[0154] The memory 804 is configured to store various types of data to support operations on the device 800. Examples of such data include instructions for any application or method operating on the device 800, contact data, phone book data, messages, pictures, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0155] The power component 806 provides power to the various components of the device 800. The power component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device 800.

[0156] The multimedia component 808 includes a screen that provides an output interface between the device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0157] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the device 800 is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 804 or sent via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

[0158] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: home button, volume button, start button, and lock button.

[0159] The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the device 800. For example, the sensor assembly 814 can detect the open / closed state of the device 800, the relative positioning of components, such as the display and keypad of the device 800, and the sensor assembly 814 can also detect the position change of the device 800 or a component of the device 800, the presence or absence of user contact with the device 800, the orientation or acceleration / deceleration of the device 800, and the temperature change of the device 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of a nearby object without any physical contact. The sensor assembly 814 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0160] The communication component 816 is configured to facilitate wired or wireless communication between the device 800 and other devices. The device 800 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0161] In an exemplary embodiment, the apparatus 800 may be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to perform the above methods.

[0162] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, and the instructions can be executed by the processor 820 of the device 800 to complete the above-mentioned automatic dialogue flow analysis method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0163] Figure 81 is a schematic diagram of the structure of the server in an embodiment of the present invention. The server 1900 may have relatively large differences due to different configurations or performances, and may include one or more central processing units (CPU) 1922 (for example, one or more processors) and memory 1932, and one or more storage media 1930 (for example, one or more mass storage devices) storing application programs 1942 or data 1944. Among them, the memory 1932 and the storage medium 1930 can be short-term storage or permanent storage. The program stored in the storage medium 1930 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the server. Furthermore, the central processing unit 1922 can be configured to communicate with the storage medium 1930 and execute a series of instruction operations in the storage medium 1930 on the server 1900.

[0164] The server 1900 may also include one or more power supplies 1926, one or more wired or wireless network interfaces 1950, one or more input and output interfaces 1958, one or more keyboards 1956, and / or, one or more operating systems 1941, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0165] A non-transitory computer-readable storage medium, when the instructions in the storage medium are executed by a processor of the device, enables the device to perform the above method.

[0166] Those skilled in the art will readily appreciate other embodiments of the invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed in this disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

[0167] It should be understood that the present invention is not limited to the exact construction that has 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 invention is limited only by the appended claims.

[0168] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for automatic analysis of dialog flow, characterized in that: The method comprises: Set multiple dialogue scenarios based on the interlocutor's identity and service standards given by the interlocutor's profession; Determine key elements of each dialogue scene and multiple different categories of tags hit by the dialogue scene; Establishing analysis models corresponding to different labels according to the key elements and their logical relationships, and obtaining an analysis engine corresponding to the identity of the interlocutor, wherein the logical relationships include at least one of a sequence relationship, an inclusion relationship, a hit relationship, a combination relationship, and a selection relationship; Obtaining the conversation data stream of each party in the conversation to be detected; Inputting the conversation data stream of each dialogue party into the analysis engine corresponding to the identity of the dialogue party, and the analysis engine identifies the key elements of the conversation data stream and determines the label information of the dialogue party based on the analysis model in the analysis engine; According to the tag information of the dialogue parties, they are classified to generate dialogue analysis results.

2. The method according to claim 1, characterized in that The dialogue parties include: the party to be inspected and the party not to be inspected; the method further includes: According to the set quality inspection rules, the conversation analysis results of the party to be quality inspected and the conversation analysis results of the party not to be quality inspected are respectively integrated to obtain a fusion result; The quality inspection evaluation result of the party to be inspected is determined according to the fusion result.

3. The method according to claim 2, characterized in that The method further comprises: The risky conversations in the conversation data stream of the party to be quality inspected are determined according to the conversation analysis results and the quality inspection evaluation results of the party to be quality inspected, and the risky conversations are marked.

4. The method according to claim 2, characterized in that: The method further comprises: Recording the conversation data flow of the party to be inspected in each conversation, as well as the conversation analysis results and quality inspection evaluation results of the party to be inspected obtained based on the conversation; Generate a quality inspection year-on-year view of the party to be inspected based on the recorded conversation analysis results and quality inspection evaluation results of the party to be inspected within a certain period; and / or A quality inspection work view is generated based on the conversation analysis results and quality inspection evaluation results of each party to be inspected obtained based on the conversations between different parties to be inspected and non-quality inspection parties with the same identity within a certain period of record.

5. The method according to claim 4, characterized in that The quality inspection month-on-month view and the quality inspection worker view are in any of the following forms: a line graph, a bar graph, or a pie chart.

6. The method according to any one of claims 2 to 5, characterized in that: The party to be inspected is the service provider; the party not to be inspected is the client; The conversation scenarios corresponding to the service provider include any one or more of the following: failure to promote business, insulting customers, failure to expand contact information, and invalid conversations; the key elements of the conversation scenarios corresponding to the service provider include any one or more of the following: key terms, banned words, selling competing products, and illegal keywords; The dialogue scenarios corresponding to the client include any one or more of the following: accepting service, being vague and hesitant, and refusing service; The key elements corresponding to the dialogue scene of the client include any one or more of the following: speaking speed, hanging up action, and tone.

7. A device for automatically analyzing a dialog flow, characterized in that: The device comprises: A data stream acquisition module, used to acquire the conversation data stream of each dialogue party in the conversation to be detected, and input the conversation data stream of each dialogue party into the analysis engine corresponding to the identity of the dialogue party; the analysis engine identifies key elements of the conversation data stream, and determines and outputs the label information of the dialogue party based on each analysis model in the analysis engine; An analysis result acquisition module is used to set multiple dialogue scenarios according to the interlocutor identities and service standards assigned by the interlocutors' occupations; determine the key elements of each dialogue scenario and multiple different categories of tags hit by the dialogue scenarios; establish analysis models corresponding to different tags based on the key elements and their logical relationships, and obtain an analysis engine corresponding to the interlocutor identities, wherein the logical relationships include at least one of a sequential relationship, an inclusion relationship, a hit relationship, a joint relationship, and a selection relationship; and determine the dialogue analysis results of the interlocutors based on the tag information of the interlocutors output by the analysis engine.

8. An electronic device, characterized in that: include: One or more processors, memory; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method according to any one of claims 1 to 6.

9. A readable storage medium having instructions stored thereon, wherein the instructions are executed to implement the method according to any one of claims 1 to 6.

Images