A legal consultation method and system based on intelligent interaction on IPTV terminal

Abstract

The present application discloses an IPTV-based legal consultation method and system based on intelligent interaction, which collects user voice signals and converts the collected voice signals into text information, obtains the legal intention of the user's consultation based on the text information, and then determines the legal category domain. According to the located legal category domain, the query statement paradigm is retrieved for multiple rounds of voice interaction to determine the legal fact information, and finally generates customized legal consultation results based on the legal fact information and feeds them back to the user using IPTV. The present application uses IPTV as the display medium for the legal consultation process and results, and cooperates with intelligent interactive devices, set-top boxes and cloud servers to realize remote legal consultation in a home environment.

Description

Technical Field

[0001] The present application relates to the field of legal consultation, and in particular to an IPTV-end legal consultation method and system based on intelligent interaction. Background Art

[0002] As a powerful weapon for modern social security to protect the legitimate rights and interests of the people, the law has been increasingly valued by the people. With the development of the economy, the speed of social development in my country has been continuously improving, and the material living standards of the people have been continuously improving. Their spiritual needs for social services, entertainment, etc. have also been continuously increasing. The process of urbanization has led to disputes and conflicts between people and between people and organizations becoming more and more complicated, and the business needs for legal consultation have also changed accordingly. With the continuous acceleration of the legalization process in my country, people's legal awareness has been continuously enhanced, which has helped the prosperity of my country's legal consultation market. Traditional legal consultation business is mainly conducted in law firms and other related places, with lawyers and clients communicating face to face. This method wastes the client's time, and for relatively backward areas, law firms are relatively scarce. The above face-to-face consultation method still has many inconveniences and cannot achieve timely and efficient legal consultation. Therefore, there is an urgent need for a remote legal consultation that can be achieved in a home environment. Summary of the invention

[0003] (I) Purpose of application

[0004] Based on this, in order to save user time, improve user experience, and realize remote legal consultation in a home environment, this application discloses the following technical solutions.

[0005] (II) Technical solution

[0006] The present application discloses an IPTV-side legal consultation method based on intelligent interaction, comprising:

[0007] Collecting user voice signals and converting the voice signals into text information;

[0008] locating a legal category domain based on the text information;

[0009] Generate legal fact information through multi-round voice interaction based on a multi-round dialogue model corresponding to the legal category domain;

[0010] A customized legal consultation result is generated according to the legal fact information, and the customized legal consultation result is fed back to the user through IPTV.

[0011] In a possible implementation manner, locating the legal category domain based on the text information includes:

[0012] Outputting the text information as the legal intent of the user's consultation using a deep learning intent recognition model;

[0013] The legal category domain is located in a legal category domain library according to the legal intent of the user consultation.

[0014] In a possible implementation, the generating legal fact information through multi-round voice interaction based on the multi-round dialogue model corresponding to the legal category domain includes:

[0015] Retrieving the inquiry statement paradigm according to the multi-round dialogue model corresponding to the legal category domain;

[0016] Replacing key inquiry information of the inquiry statement paradigm with user expressions, and performing multiple rounds of voice interactions based on the user expressions;

[0017] User input information is collected based on the multiple rounds of voice interaction, and variables are assigned to the user input information to generate legal fact information.

[0018] In a possible implementation, the method further includes: interactively determining legal fact information through IPTV, specifically including:

[0019] The IPTV displays the inquiry statement paradigm and provides corresponding answer options, and the user answers through human-computer interaction to determine the legal fact information;

[0020] Among them, the human-computer interaction method includes physical button interaction, voice interaction or touch interaction.

[0021] In a possible implementation, generating a customized legal consultation result according to the legal fact information, and feeding back the customized legal consultation result to the user through IPTV, includes:

[0022] Performing reasoning and decision calculation on the legal fact information based on the reasoning and decision strategy model;

[0023] Generate customized legal consulting results using a format document generation tool based on the results of the reasoning decision calculation;

[0024] Feedback the customized legal consultation results to the user via IPTV;

[0025] The information fed back to the user via IPTV is in the form of at least one of voice broadcast, animation broadcast, and text display.

[0026] As a second aspect of the present application, the present application also discloses an IPTV-end legal consultation system based on intelligent interaction, comprising:

[0027] Intelligent interactive device, used to collect user voice signals and convert the voice signals into text information, and perform human-computer interaction with the user;

[0028] A cloud server, used to obtain legal fact information and generate customized legal consultation results based on the legal fact information;

[0029] Set-top box, used to realize communication between IPTV and cloud server and intelligent interactive devices;

[0030] IPTV is used to broadcast and / or display interactive content and consultation results.

[0031] In a possible implementation, the cloud server includes:

[0032] A category domain locating module, used to locate the legal category domain based on the text information;

[0033] A legal fact generation module, used for performing multi-round voice interaction based on a multi-round dialogue model corresponding to the legal category domain to generate legal fact information;

[0034] The result generation module is used to generate customized legal consultation results based on the legal fact information.

[0035] In a possible implementation, the category domain positioning module includes:

[0036] A legal intent determination unit, configured to output the text information as the legal intent consulted by the user using a deep learning intent recognition model;

[0037] A category domain locating unit is used to locate the legal category domain in the legal category domain library according to the legal intent consulted by the user.

[0038] In a possible implementation, the legal fact generation module includes:

[0039] An inquiry statement retrieving unit, used to retrieve an inquiry statement paradigm according to a multi-round dialogue model corresponding to the legal category domain;

[0040] A voice interaction unit, used to replace key inquiry information of the inquiry statement paradigm with user expressions, and perform multiple rounds of voice interactions based on the user expressions;

[0041] A legal fact generation unit is used to collect user input information based on the multiple rounds of voice interaction, and to perform variable assignment on the user input information to generate legal fact information.

[0042] In a possible implementation, the result generation module includes:

[0043] An inference decision calculation unit, used for performing inference decision calculation on the legal fact information based on the inference decision strategy model;

[0044] A result generating unit is used to generate customized legal consulting results based on the result of the reasoning decision calculation using a format document generating tool.

[0045] (III) Beneficial effects

[0046] The present application discloses an IPTV-end legal consultation method and system based on intelligent interaction. By providing legal consultation on an IPTV-end based on an intelligent interactive device, the user's time is saved, the user experience is improved, and remote legal consultation is realized in a home environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain and illustrate the present application, but should not be construed as limiting the scope of protection of the present application.

[0048] Figure 1 It is a flowchart of an IPTV-side legal consultation method based on intelligent interaction disclosed in this application.

[0049] Figure 2 This is a structural block diagram of an IPTV-side legal consultation system based on intelligent interaction disclosed in this application. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the implementation of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below in conjunction with the drawings in the embodiments of this application.

[0051] Reference below Figure 1 The present invention discloses an IPTV-based legal consultation method embodiment based on intelligent interaction. Figure 1 As shown, the method disclosed in this embodiment mainly includes the following steps 100 to 400.

[0052] Step 100: Collect user voice signals and convert the voice signals into text information.

[0053] Specifically, user voice signals are collected through intelligent interactive devices, and the intelligent interactive devices include devices with voice collection and voice recognition functions, including smart speakers, smart phones and other smart appliances. The intelligent interactive devices have a built-in voice recognition module, and the voice recognition module converts the collected voice signals into text information. Furthermore, step 100 includes steps 110 to 130.

[0054] Step 110: pre-process the collected speech signal.

[0055] Specifically, the voice signal collected by the intelligent interactive device needs to be preprocessed to obtain feature parameters, so as to compare the parameters in the template parameter library with the recognition results in the subsequent template matching and obtain text information. Further, step 110 also includes steps 111 to 114.

[0056] Step 111: pre-filter the speech signal.

[0057] Specifically, the bandpass filter is used as a pre-filter of the speech signal to pre-filter the speech signal, determine the upper and lower cut-off frequencies, and quantize the speech signal to weaken some spectral components, prevent aliasing interference and suppress power supply frequency interference.

[0058] Step 112: pre-emphasize the speech signal.

[0059] Specifically, before feature extraction through a digital filter, the speech signal needs to be pre-emphasized, that is, the high-frequency spectrum is enhanced and the connection section between the high-frequency part and the low-frequency part is smoothed to solve the spectrum under the same signal-to-noise ratio condition, making subsequent analysis more convenient and quick.

[0060] Step 113, frame-by-frame windowing of the speech signal to obtain a short-term stationary signal of the speech.

[0061] Specifically, firstly, a suitable window shape and window length are selected, and the speech signal is subjected to frame-by-frame windowing processing by sliding the window function to obtain the short-term stability of the speech signal.

[0062] Step 114: perform endpoint detection on the speech signal to obtain the start and end points of the target speech signal.

[0063] Specifically, firstly, a short-time energy analysis is performed on the speech signal to obtain a detailed description of the amplitude change of the speech signal, then a short-time average zero-crossing rate analysis is performed on the speech signal to obtain the frequency information of the speech signal, and finally, endpoint detection is performed using a double threshold endpoint detection method by combining the two parameters of short-time energy and short-time zero-crossing rate to obtain the start and end points of the target speech signal.

[0064] Step 120: extracting feature parameters from the preprocessed speech signal.

[0065] Specifically, the characteristic parameters include parameters such as pitch period and resonance peak. The most commonly used comprehensive characteristic parameters are: Linear Prediction Cepstral Coefficient (LPCC) and Mel Cepstral Coefficient (MFCC). The two characteristic parameters operate on the speech signal in the cepstral domain. Preferably, the Mel Cepstral Coefficient is used as the characteristic parameter of the speech signal. The Mel Cepstral Coefficient (MFCC) is the most widely used speech characteristic parameter in the speech recognition system. After the input speech signal is pre-emphasized, framed and windowed, the spectrum of the signal is obtained by fast Fourier transform, the signal spectrum is passed through the Mel filter group, the logarithmic energy of each filter group is output, the logarithmic spectrum is obtained, and finally the logarithmic spectrum is subjected to discrete cosine transform DCT to obtain the Mel Cepstral Coefficient (MFCC) coefficient.

[0066] Step 130: Obtain the speech signal recognition result.

[0067] Specifically, a pre-trained model is used for speech signal recognition, preferably, an HMM model is selected, modeling is performed according to a continuous HMM model, and the HMM model is trained, speech recognition is performed using the trained HMM model according to the extracted feature parameters, and a recognition result is obtained, that is, the speech signal is converted into text information.

[0068] Step 200, locating the legal category domain based on the text information.

[0069] Specifically, after obtaining the text information in step 100, the intelligent interactive device transmits the text information to the cloud server, and the cloud server processes the text information. Further, step 200 also includes steps 210 to 220.

[0070] Step 210, using a deep learning intention recognition model to output the text information as the legal intention of the user's consultation;

[0071] Specifically, to obtain the legal intent of the user's consultation, that is, to determine the user's purpose of legal consultation, first extract the keywords in the text information as classification features, and classify the user input information according to the classification features using a deep learning intention recognition model to obtain the legal intent of the user's consultation. According to the characteristics of legal consultation, preferably, the legal intent of the user's consultation may be divorce, resolution of labor disputes, etc.

[0072] Step 220: locate the legal category domain in the legal category domain library according to the legal intent consulted by the user.

[0073] Specifically, the cloud server is provided with a legal category domain library. The legal category domain library has different built-in legal category domains. After obtaining the user's legal intent, the legal category domain is located in the legal category domain library according to the user's legal intent. Preferably, the legal category domain includes: civil consultation domain, criminal consultation domain, foreign-related consultation domain, corporate legal issue consultation domain, etc., and each domain also includes subdomains. For example, the civil consultation domain includes a demolition and resettlement consultation subdomain, a creditor's rights and debts consultation subdomain, a medical dispute consultation subdomain, a traffic accident responsibility consultation subdomain, a marriage consultation subdomain, etc., to perform preliminary category domain division of user legal consultation. The legal category domain here determines the human-computer dialogue mechanism applicable in subsequent multiple rounds of voice interaction, including the inquiry statements and user expressions used in the human-computer dialogue.

[0074] Step 300: Perform multiple rounds of voice interaction based on the multiple rounds of dialogue model corresponding to the legal category domain to generate legal fact information.

[0075] Specifically, each legal category domain corresponds to a multi-round dialogue model, and human-computer dialogue content is generated based on the corresponding multi-round dialogue model. The intelligent interactive device conducts dialogue with the user to obtain user input information. Further, step 300 also includes steps 310 to 330.

[0076] Step 310: retrieve the inquiry statement paradigm according to the multi-round dialogue model corresponding to the legal category domain.

[0077] Specifically, the legal category domain is disassembled, and the key elements of a large number of cases and expert experience are captured and summarized based on machine learning. The multi-round dialogue model dialogue library is automatically generated through text. Based on the multi-round dialogue model, the corresponding query statement paradigm in the multi-round dialogue model dialogue library is retrieved to guide users to respond and improve the input legal information, so as to output accurate consultation results.

[0078] Step 320: Replace the key inquiry information of the inquiry statement paradigm with user expressions, and perform multiple rounds of voice interaction based on the user expressions.

[0079] Specifically, part of the inquiry information in the inquiry statement paradigm retrieved in step 310 is replaced with user expressions to prevent the professionalism and complexity of legal terms from making it difficult for users to understand. Replacing with user expressions can help users understand the content of the conversation and give more accurate responses.

[0080] Step 330: Collect user input information based on the multiple rounds of voice interaction, and assign variables to the user input information to generate legal fact information.

[0081] Specifically, multiple rounds of voice interaction are performed based on the generated multi-round conversation model, user input information is collected during each round of voice interaction, and the user input information is assigned to legal feature variables to generate a set of legal feature variables, namely the legal fact information.

[0082] In at least one embodiment, step 300 further includes step 340, interactively determining legal fact information through IPTV, specifically including:

[0083] The IPTV displays the inquiry statement paradigm and provides corresponding answer options, and the user answers through human-computer interaction to determine the legal fact information;

[0084] Among them, the human-computer interaction method includes physical button interaction, voice interaction or touch interaction.

[0085] Specifically, when the user of the intelligent interactive device performs human-computer interaction, the cloud server simultaneously transmits the interactive content to the IPTV through the set-top box. The IPTV screen can display the query statement, and the answer options of the query statement are attached. The user can use physical buttons such as remote control buttons, voice interaction such as direct interaction through the intelligent speaker, or touch interaction such as direct selection of the answer option through the IPTV screen. Multiple rounds of interaction can be completed by the cloud server in conjunction with the intelligent voice device or IPTV alone, or by the cloud server in conjunction with the intelligent voice device and IPTV.

[0086] In this embodiment, the cloud server is connected to the set-top box and the intelligent interactive device through a local area network. The set-top box controls the IPTV as a display medium. The IPTV and the intelligent interactive device are also connected through the set-top box. The intelligent interactive device interacts with the set-top box through a local area network or Bluetooth. The IPTV and the intelligent interactive device cooperate with each other to realize the visualization and diversification of intelligent interaction, improve the interaction efficiency, bring convenience to users, and greatly improve the user experience.

[0087] Step 400, generating a customized legal consultation result based on the legal fact information, and feeding back the customized legal consultation result to the user through IPTV.

[0088] Specifically, after obtaining the legal fact information composed of the set of legal feature variables, a customized legal consultation result is generated based on the legal fact information, and the legal consultation result is displayed through the IPTV screen. Further, step 400 includes steps 410 to 430.

[0089] Step 410: Perform reasoning and decision calculation on the legal fact information based on the reasoning and decision strategy model.

[0090] Specifically, after obtaining the legal fact information composed of the set of legal characteristic variables, reasoning and decision calculations are performed based on different dimensions for the assigned legal characteristic variables, and the dimensions include at least one of the logical relationship between legal characteristic values, the corresponding relationship between legal characteristic values ​​and results, and support for the calculation of complex expressions. The logical relationship between the legal characteristic values ​​is, for example, if variable A has been assigned a value of X, then variable B is automatically assigned a value of Y, and variable C will not be assigned a value again; the corresponding relationship between the legal characteristic values ​​and the result is, for example, if variable A is assigned a value of X, then the result is displayed as XXX; the calculation of complex expressions is to indicate the calculation of complex expressions combined with addition, subtraction, multiplication, division, judgment, and loop.

[0091] Step 420, using a format document generation tool to generate customized legal consulting results based on the results of the reasoning decision calculation.

[0092] Specifically, firstly, a standard document template in word format is uploaded to the cloud server, and based on the standard document template and the inference decision calculation result of step 410, a customized legal consultation result in a standard format is automatically generated.

[0093] Step 430: Feedback the customized legal consultation results to the user via IPTV.

[0094] Specifically, the cloud server transmits the customized legal consultation result to the IPTV through the set-top box, and the IPTV feeds back the customized legal consultation result to the user through the IPTV.

[0095] It should be noted that the form of the information fed back to the user via IPTV is at least one of voice broadcast, animation broadcast, and text display.

[0096] Specifically, IPTV can display the consultation results according to user needs. The display methods include: voice broadcast, text display, animation demonstration, etc., and multiple demonstration methods can be performed simultaneously, for example: voice broadcast combined with text display to provide feedback on legal consultation results.

[0097] In this embodiment, IPTV is used to realize a variety of display methods of legal consultation results, thereby improving user experience. Using IPTV as a display medium to realize remote legal consultation in a home environment is a brand-new business model.

[0098] In at least one embodiment, after the legal consultation result is generated, if the user still has questions or doubts about the legal consultation result, manual consultation can be conducted, which is real-time online consultation with a cooperating law firm or other personnel who can provide a legal consultation venue. Further, this embodiment is implemented through steps 510 to 520.

[0099] Step 510: The legal fact information and the legal category domain are transmitted to a legal partner. The legal partner selects lawyers who can provide services and feeds back the lawyer information on the IPTV.

[0100] Specifically, the legal fact information is transmitted to the legal partner, which is a cooperating law firm or other place that can provide legal consultation. Lawyers in the corresponding field are screened according to the legal category domain, and further screening is performed based on the real-time work status of lawyers who meet the requirements. Lawyers who can provide legal consultation services are screened out and the lawyer information is fed back on IPTV. The lawyer information includes basic identity information, personal classic cases, personal expertise, personal and company ratings, etc.

[0101] Step 520: The user independently selects a lawyer for consultation through the lawyer information displayed on the IPTV.

[0102] Specifically, users can make their own choices based on the lawyer information displayed on IPTV. The system can also randomly match lawyers for legal consulting services. After selecting a lawyer for legal consulting services, the user can connect with the lawyer through IPTV, and the connection method can be a video call or a voice call.

[0103] Furthermore, if there is no matching lawyer at present, that is, all lawyers in the field who meet the requirements are busy, an appointment service can be provided to coordinate the time between the lawyer and the user to make an appointment for legal consultation services, and then provide legal consultation according to the appointment time.

[0104] In at least one embodiment, an automatic recording function is also included, that is,

[0105] IPTV automatically records the legal consultation process and results and generates legal consultation record notes.

[0106] Specifically, IPTV automatically records the legal consultation process and results, and generates legal consultation notes. At the same time, it can also realize automatic recording and screen recording during manual consultation. After the consultation, the user can check the legal consultation notes at any time, listen to the recordings repeatedly, watch the screen recordings, etc., to improve the user experience and upgrade the quality of legal consultation.

[0107] In at least one embodiment, it also has an evaluation and recommendation function, and further includes steps 610 to 630.

[0108] Step 610: The user evaluates the lawyer's services using IPTV and / or smart interactive devices.

[0109] Specifically, after completing a legal consultation, the user can evaluate the lawyer's services through IPTV and / or smart interactive devices. The rating content includes the lawyer's service attitude, professionalism, and level of response.

[0110] Step 620: Generate a score display system based on the user evaluation.

[0111] Specifically, a score display system is generated based on the evaluations of all users, and the score display system includes the overall score and ranking of the law firm, the score and ranking of each lawyer in the law firm, and the score and ranking of lawyers in each legal field.

[0112] Step 630: recommend lawyers and law firms to the user based on the partner score display system.

[0113] Specifically, the system recommends lawyers and law firms to users based on the partner score display system, making it easier for users to choose law firms and lawyers, thereby bringing a better user experience.

[0114] Reference below Figure 2 The present invention discloses an IPTV legal consultation system based on intelligent interaction. Figure 2 As shown, the system disclosed in this embodiment includes:

[0115] Intelligent interactive device, used to collect user voice signals and convert the voice signals into text information, and perform human-computer interaction with the user;

[0116] A cloud server, used to obtain legal fact information and generate customized legal consultation results based on the legal fact information;

[0117] Set-top box, used to realize communication between IPTV and cloud server and intelligent interactive devices;

[0118] IPTV is used to broadcast and / or display interactive content and consultation results.

[0119] In at least one embodiment, the cloud server includes:

[0120] A category domain locating module, used to locate the legal category domain based on the text information;

[0121] A legal fact generation module, used for performing multi-round voice interaction based on a multi-round dialogue model corresponding to the legal category domain to generate legal fact information;

[0122] The result generation module is used to generate customized legal consultation results based on the legal fact information.

[0123] In at least one embodiment, the category domain positioning module includes:

[0124] A legal intent determination unit, configured to output the text information as the legal intent consulted by the user using a deep learning intent recognition model;

[0125] A category domain locating unit is used to locate the legal category domain in the legal category domain library according to the legal intent consulted by the user.

[0126] In at least one embodiment, the legal fact generation module includes:

[0127] An inquiry statement retrieving unit, used to retrieve an inquiry statement paradigm according to a multi-round dialogue model corresponding to the legal category domain;

[0128] A voice interaction unit, used to replace key inquiry information of the inquiry statement paradigm with user expressions, and perform multiple rounds of voice interactions based on the user expressions;

[0129] A legal fact generation unit is used to collect user input information based on the multiple rounds of voice interaction, and to perform variable assignment on the user input information to generate legal fact information.

[0130] In at least one embodiment, the result generation module includes:

[0131] An inference decision calculation unit, used for performing inference decision calculation on the legal fact information based on the inference decision strategy model;

[0132] A result generating unit is used to generate customized legal consulting results based on the result of the reasoning decision calculation using a format document generating tool.

[0133] In at least one embodiment, further comprising:

[0134] An IPTV interactive module is used for displaying the inquiry statement paradigm on IPTV with corresponding answer options, and the user answers through human-computer interaction to determine the legal fact information;

[0135] Among them, the human-computer interaction method includes physical button interaction, voice interaction or touch interaction.

[0136] In at least one embodiment, the feedback form of the IPTV when feeding back the legal consultation results includes: at least one of voice broadcast, animation broadcast or text display.

[0137] The division of modules and units in this article is only a division of logical functions. In actual implementation, there may be other division methods, for example, multiple modules and units can be combined or integrated into another system. The modules and units described as separate components can be physically separated or inseparable. The components displayed as units can be physical units or not, that is, they can be located in a specific place or distributed in grid units. Therefore, some or all of the units can be selected according to actual needs to implement the scheme of the embodiment.

[0138] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. An IPTV-based legal consultation method based on intelligent interaction, characterized in that: include: Collecting user voice signals and converting the voice signals into text information; Collecting user voice signals through an intelligent interactive device, wherein the intelligent interactive device includes a device with voice collection and voice recognition functions, and the intelligent interactive device has a built-in voice recognition module, and the voice recognition module converts the collected voice signals into text information; The specific steps of converting speech signals into text information include: Pre-filtering the speech signal; using a bandpass filter as a pre-filter of the speech signal, pre-filtering the speech signal, determining its upper and lower cut-off frequencies, and quantizing the speech signal to weaken some spectral components to prevent aliasing interference and suppress power supply frequency interference; Pre-emphasize the speech signal; The speech signal is framed and windowed to obtain a short-term stable signal of the speech; a suitable window shape and window length are selected, and the speech signal is framed and windowed by sliding the window function to obtain the short-term stability of the speech signal; Performing endpoint detection on the speech signal to obtain the start and end points of the speech signal; performing short-time energy analysis on the speech signal to obtain a detailed description of the amplitude change of the speech signal; performing short-time average zero-crossing rate analysis on the speech signal to obtain the frequency information of the speech signal; performing endpoint detection using a double threshold endpoint detection method based on the two parameters of short-time energy and short-time zero-crossing rate to obtain the start and end points of the speech signal; Extract characteristic parameters of the preprocessed speech signal; the characteristic parameters include pitch period and resonance peak; after pre-emphasis and frame windowing of the input speech signal, obtain the spectrum of the signal through fast Fourier transform, pass the signal spectrum through the Mel filter group, output the logarithmic energy of each filter group, obtain the logarithmic spectrum, and finally obtain the Mel cepstral coefficient coefficient through discrete cosine transform DCT of the logarithmic spectrum; Acquire a speech signal recognition result; perform modeling according to a continuous HMM model, perform model training on the HMM model, perform speech recognition using the trained HMM model according to the extracted feature parameters, and obtain a recognition result, that is, convert the speech signal into text information; The legal category domain is located based on the text information, specifically including: extracting keywords in the text information as classification features, classifying the user input information using a deep learning intention recognition model according to the classification features, obtaining the legal intention of the user consultation, and locating the legal category domain in a legal category domain library according to the legal intention of the user consultation, wherein the legal category domain library has different legal category domains built in, and each legal category domain also includes subdomains; the legal category domain determines the human-computer dialogue mechanism applicable in subsequent multiple rounds of voice interaction, and the human-computer dialogue mechanism includes the query statements and user expressions used in the human-computer dialogue; Based on the multi-round dialogue model corresponding to the legal category domain, the user is guided to answer from the corresponding inquiry statement paradigm in the corresponding multi-round dialogue model dialogue library, the input legal information is improved, the key inquiry information of the inquiry statement paradigm is replaced with the user expression, and multi-round voice interaction is performed based on the user expression, the user input information is collected based on the multi-round voice interaction, and the user input information is assigned variables to generate a set of legal feature variables as legal fact information; wherein each legal category domain corresponds to a multi-round dialogue model; wherein the inquiry statement paradigm is displayed through IPTV with corresponding answer options, and the user answers through human-computer interaction, thereby determining the legal fact information; Interactively determine legal facts through IPTV, including: The IPTV displays the inquiry statement paradigm and the corresponding answer options, and the user answers through human-computer interaction to determine the legal fact information; wherein the human-computer interaction method includes physical button interaction, voice interaction or touch interaction; When the user of the intelligent interactive device performs human-computer interaction, the cloud server transmits the interactive content to the IPTV through the set-top box at the same time. The IPTV screen displays the query statement and the answer options of the query statement. The user directly selects the answer option by using physical buttons, voice interaction or touch interaction. Multiple rounds of interaction are completed by the cloud server in cooperation with the intelligent voice device or IPTV alone, or by the cloud server in cooperation with the intelligent voice device and IPTV together. Obtain legal fact information composed of the set of legal feature variables, and based on the inference decision strategy model, perform inference decision calculations based on different dimensions for the assigned legal feature variables, wherein the dimensions include at least one of the logical relationship between legal feature values, the corresponding relationship between legal feature values ​​and results, and the calculation of supporting complex expressions; generate customized legal consulting results based on the results of the inference decision calculations using a format document generation tool; and feed back the customized legal consulting results to the user via IPTV; wherein the form of the feedback to the user via IPTV is at least one of voice broadcast, animation broadcast, and text display; Based on the results of the reasoning and decision calculation, a format document generation tool is used to generate customized legal consulting results, including: First, a standard document template in word format is uploaded to the cloud server, and customized legal consultation results in a standard format are automatically generated based on the standard document template and the inference decision calculation results; After the legal consultation results are generated, if the user still has questions or doubts about the legal consultation results, he or she can also conduct manual consultation, which is real-time online consultation with the cooperating law firm or other personnel who can provide legal consultation venues, including: The legal fact information and the legal category domain are transmitted to the legal partner, and the legal partner selects lawyers who can provide services and feeds back the lawyer information on the IPTV; the legal fact information is transmitted to the legal partner, and the legal partner is a cooperating law firm or other place that can provide legal consultation, and lawyers in corresponding fields are screened according to the legal category domain, and further screening is performed according to the real-time working status of lawyers who meet the requirements, and lawyers who can provide legal consultation services are screened and the lawyer information is fed back on the IPTV, and the lawyer information includes basic identity information, personal classic cases, personal expertise, and personal and company ratings; Users can choose lawyers for consultation based on the lawyer information displayed on IPTV; Users can make their own choices based on the lawyer information displayed on IPTV, or randomly match lawyers for legal consulting services. After selecting a lawyer for legal consulting services, they can connect with the lawyer through IPTV in the form of video or voice calls; If there is no matching lawyer at present, that is, all the lawyers in the field who meet the requirements are busy, the appointment service will be used to coordinate the lawyer and the user to make an appointment for legal consultation services, and then the legal consultation will be carried out according to the appointment time; Use IPTV to automatically record the legal consultation process and results, and generate legal consultation record notes; After completing a legal consultation, users can evaluate the lawyer's service through IPTV and / or smart interactive devices. The rating includes the lawyer's service attitude, professionalism, and answer level. Generate a score display system based on user reviews; Among them, a score display system is generated based on the evaluations of all users, and the score display system includes the overall score and ranking of the law firm, the score and ranking of each lawyer in the law firm, and the score and ranking of lawyers in each legal field; Recommending lawyers and law firms to users based on the partner score display system; Among them, the system recommends lawyers and law firms to users based on the partner score display system.

2. An IPTV legal consultation system based on intelligent interaction, characterized in that: include: Intelligent interactive device, used to collect user voice signals and convert the voice signals into text information, and perform human-computer interaction with the user; A cloud server, used to obtain legal fact information and generate customized legal consultation results based on the legal fact information; Set-top box, used to realize communication between IPTV and cloud server and intelligent interactive devices; IPTV, used to broadcast and / or display interactive content and consultation results; The cloud server comprises: A category domain positioning module is used to locate the legal category domain based on the text information; the category domain positioning module includes: a legal intent determination unit, which is used to extract keywords in the text information as classification features, and classify the user input information using a deep learning intent recognition model according to the classification features to obtain the legal intent of the user consultation; a category domain positioning unit is used to locate the legal category domain in a legal category domain library according to the legal intent of the user consultation, the legal category domain library has different legal category domains built in, and each legal category domain also includes subdomains; the legal category domain determines the human-computer dialogue mechanism applicable in subsequent multiple rounds of voice interaction, and the human-computer dialogue mechanism includes the query statements and user expressions used in the human-computer dialogue; A legal fact generation module, used for performing multi-round voice interaction based on a multi-round dialogue model corresponding to the legal category domain to generate legal fact information; A result generation module, used to generate customized legal consultation results based on the legal fact information; The legal fact generation module includes: An inquiry statement retrieval unit, used to retrieve an inquiry statement paradigm according to the multi-round dialogue model corresponding to the legal category domain, guide the user to answer, and complete the input legal information; A voice interaction unit, used to replace key inquiry information of the inquiry statement paradigm with user expressions, and perform multiple rounds of voice interactions based on the user expressions; A legal fact generation unit is used to collect user input information based on the multi-round voice interaction, and assign variables to the user input information to generate a set of legal feature variables as legal fact information; wherein each legal category domain corresponds to a multi-round dialogue model; wherein the inquiry statement paradigm is displayed through IPTV with corresponding answer options, and the user answers through human-computer interaction, thereby determining the legal fact information; The result generation module comprises: An inference decision calculation unit, used to obtain the legal fact information composed of the set of legal feature variables, and based on the inference decision strategy model, perform inference decision calculations on the assigned legal feature variables based on different dimensions, wherein the dimensions include at least one of the logical relationship between legal feature values, the corresponding relationship between legal feature values ​​and results, and the calculation supporting complex expressions; A result generating unit, used to generate a customized legal consultation result based on the result of the reasoning decision calculation using a format document generating tool; and feed back the customized legal consultation result to the user via IPTV; wherein the form of the feedback to the user via IPTV is at least one of voice broadcast, animation broadcast, and text display; Collecting user voice signals through an intelligent interactive device, wherein the intelligent interactive device includes a device with voice collection and voice recognition functions, and the intelligent interactive device has a built-in voice recognition module, and the voice recognition module converts the collected voice signals into text information; The specific steps of converting speech signals into text information include: Pre-filtering the speech signal; using a bandpass filter as a pre-filter of the speech signal, pre-filtering the speech signal, determining its upper and lower cut-off frequencies, and quantizing the speech signal to weaken some spectral components to prevent aliasing interference and suppress power supply frequency interference; Pre-emphasize the speech signal; The speech signal is framed and windowed to obtain a short-term stable signal of the speech; a suitable window shape and window length are selected, and the speech signal is framed and windowed by sliding the window function to obtain the short-term stability of the speech signal; Performing endpoint detection on the speech signal to obtain the start and end points of the speech signal; performing short-time energy analysis on the speech signal to obtain a detailed description of the amplitude change of the speech signal; performing short-time average zero-crossing rate analysis on the speech signal to obtain the frequency information of the speech signal; performing endpoint detection using a double threshold endpoint detection method based on the two parameters of short-time energy and short-time zero-crossing rate to obtain the start and end points of the speech signal; Extract characteristic parameters of the preprocessed speech signal; the characteristic parameters include pitch period and resonance peak; after pre-emphasis and frame windowing of the input speech signal, obtain the spectrum of the signal through fast Fourier transform, pass the signal spectrum through the Mel filter group, output the logarithmic energy of each filter group, obtain the logarithmic spectrum, and finally obtain the Mel cepstral coefficient coefficient through discrete cosine transform DCT of the logarithmic spectrum; Acquire a speech signal recognition result; perform modeling according to a continuous HMM model, perform model training on the HMM model, perform speech recognition using the trained HMM model according to the extracted feature parameters, and obtain a recognition result, that is, convert the speech signal into text information; Interactively determine legal facts through IPTV, including: The IPTV displays the inquiry statement paradigm and the corresponding answer options, and the user answers through human-computer interaction to determine the legal fact information; wherein the human-computer interaction method includes physical button interaction, voice interaction or touch interaction; When the user of the intelligent interactive device performs human-computer interaction, the cloud server transmits the interactive content to the IPTV through the set-top box at the same time. The IPTV screen displays the query statement and the answer options of the query statement. The user directly selects the answer option by using physical buttons, voice interaction or touch interaction. Multiple rounds of interaction are completed by the cloud server in cooperation with the intelligent voice device or IPTV alone, or by the cloud server in cooperation with the intelligent voice device and IPTV together. Based on the results of the reasoning and decision calculation, a format document generation tool is used to generate customized legal consulting results, including: First, a standard document template in word format is uploaded to the cloud server, and customized legal consultation results in a standard format are automatically generated based on the standard document template and the inference decision calculation results; After the legal consultation results are generated, if the user still has questions or doubts about the legal consultation results, he or she can also conduct manual consultation, which is real-time online consultation with the cooperating law firm or other personnel who can provide legal consultation venues, including: The legal fact information and the legal category domain are transmitted to the legal partner, and the legal partner selects lawyers who can provide services and feeds back the lawyer information on the IPTV; the legal fact information is transmitted to the legal partner, and the legal partner is a cooperating law firm or other place that can provide legal consultation, and lawyers in corresponding fields are screened according to the legal category domain, and further screening is performed according to the real-time working status of lawyers who meet the requirements, and lawyers who can provide legal consultation services are screened and the lawyer information is fed back on the IPTV, and the lawyer information includes basic identity information, personal classic cases, personal expertise, and personal and company ratings; Users can choose lawyers for consultation based on the lawyer information displayed on IPTV; Users can make their own choices based on the lawyer information displayed on IPTV, or randomly match lawyers for legal consulting services. After selecting a lawyer for legal consulting services, they can connect with the lawyer through IPTV in the form of video or voice calls; If there is no matching lawyer at present, that is, all the lawyers in the field who meet the requirements are busy, the appointment service will be used to coordinate the lawyer and the user to make an appointment for legal consultation services, and then the legal consultation will be carried out according to the appointment time; Use IPTV to automatically record the legal consultation process and results, and generate legal consultation record notes; After completing a legal consultation, users can evaluate the lawyer's service through IPTV and / or smart interactive devices. The rating includes the lawyer's service attitude, professionalism, and answer level. Generate a score display system based on user reviews; Among them, a score display system is generated based on the evaluations of all users, and the score display system includes the overall score and ranking of the law firm, the score and ranking of each lawyer in the law firm, and the score and ranking of lawyers in each legal field; Recommending lawyers and law firms to users based on the partner score display system; Among them, the system recommends lawyers and law firms to users based on the partner score display system.

Images