CN114817488A - Information processing method and device in live broadcast, electronic equipment and storage medium - Google Patents
Information processing method and device in live broadcast, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN114817488A CN114817488A CN202110127263.9A CN202110127263A CN114817488A CN 114817488 A CN114817488 A CN 114817488A CN 202110127263 A CN202110127263 A CN 202110127263A CN 114817488 A CN114817488 A CN 114817488A
- Authority
- CN
- China
- Prior art keywords
- question
- answer
- live broadcast
- vector
- answers
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003860 storage Methods 0.000 title claims abstract description 16
- 230000010365 information processing Effects 0.000 title claims description 38
- 238000003672 processing method Methods 0.000 title claims description 27
- 238000000034 method Methods 0.000 claims abstract description 59
- 238000012545 processing Methods 0.000 claims abstract description 30
- 230000002452 interceptive effect Effects 0.000 claims abstract description 21
- 230000008569 process Effects 0.000 claims abstract description 19
- 239000013598 vector Substances 0.000 claims description 198
- 238000010801 machine learning Methods 0.000 claims description 44
- 230000015654 memory Effects 0.000 claims description 24
- 238000004458 analytical method Methods 0.000 claims description 12
- 230000003993 interaction Effects 0.000 claims description 10
- 230000004044 response Effects 0.000 claims description 7
- 230000002860 competitive effect Effects 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 20
- 238000013473 artificial intelligence Methods 0.000 abstract description 9
- 238000003058 natural language processing Methods 0.000 abstract description 7
- 238000010586 diagram Methods 0.000 description 16
- 230000006870 function Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 7
- 238000012549 training Methods 0.000 description 7
- 238000004590 computer program Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 241000282414 Homo sapiens Species 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 235000013312 flour Nutrition 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000013515 script Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000014509 gene expression Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 241000282412 Homo Species 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 235000013601 eggs Nutrition 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Human Computer Interaction (AREA)
- Databases & Information Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application provides a method and a device for processing information in live broadcast, electronic equipment and a computer readable storage medium; relates to natural language processing technology in the field of artificial intelligence; the method comprises the following steps: presenting live content of a live room; acquiring knowledge information from an interactive event occurring in the live broadcast room in the process of presenting the live broadcast content; and responding to a question submitting operation aiming at the live broadcast room, presenting the question submitted by the question submitting operation, acquiring an answer of the question based on the knowledge information, and presenting the answer. Through the method and the device, the efficiency and the accuracy of answer to the question can be improved in the live broadcasting process.
Description
Technical Field
The present disclosure relates to internet technologies and artificial intelligence technologies, and in particular, to a method and an apparatus for processing information in live broadcast, an electronic device, and a computer-readable storage medium.
Background
Artificial Intelligence (AI) is a theory, method and technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human Intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results. As artificial intelligence technology has been researched and developed, artificial intelligence technology has been developed and applied in various fields.
Taking a live broadcast scene as an example, when watching the live broadcast content interested by audiences in a live broadcast room, the audiences can put forward problems to the anchor broadcast through the input box, and the answering efficiency aiming at the problems in the related technology is low, so that the information transmission effect of the live broadcast is influenced. For this reason, the related art has not yet made an effective solution.
Disclosure of Invention
The embodiment of the application provides an information processing method and device in live broadcasting, electronic equipment and a computer readable storage medium, which can improve the efficiency and accuracy of answer making for questions in the live broadcasting process.
The technical scheme of the embodiment of the application is realized as follows:
the embodiment of the application provides an information processing method in live broadcasting, which comprises the following steps:
presenting live content of a live room;
acquiring knowledge information from an interactive event occurring in the live broadcast room in the process of presenting the live broadcast content;
in response to a question submission operation directed to the live broadcast room, presenting a question submitted by the question submission operation, and
and acquiring an answer of the question based on the knowledge information, and presenting the answer.
In the above scheme, after presenting the answer, the method further comprises:
responding to the feedback operation aiming at the answer, and acquiring feedback information submitted by the feedback operation;
when the feedback information is of a negative type, sending the question to a client logged in with a main broadcasting account for presentation;
responding to an answer operation aiming at the question, acquiring an answer submitted by the answer operation, and presenting the submitted answer;
and forming question-answer pairs by the questions and the submitted answers, and synchronizing the question-answer pairs to a question-answer library.
An embodiment of the present application provides an information processing apparatus in live broadcasting, including:
the display module is used for presenting the live broadcast content of the live broadcast room;
the acquisition module is used for acquiring knowledge information from an interactive event occurring in the live broadcast room in the process of presenting the live broadcast content;
and the answer module is used for responding to the question submitting operation aiming at the live broadcast room, presenting the question submitted by the question submitting operation, acquiring the answer of the question based on the knowledge information and presenting the answer.
In the above scheme, the obtaining module is further configured to obtain knowledge information in real time from an interactive event occurring in the live broadcast room and/or the associated live broadcast room; wherein the interaction event comprises at least one of: displaying recommendation information, question and answer events and teaching events; wherein the associated live room comprises at least one of: the live broadcast room comprises a live broadcast room having a competitive relationship with the live broadcast room, a live broadcast room of a main broadcast having a social relationship with a main broadcast in the live broadcast room, and a live broadcast room of the same type as the live broadcast room.
In the above scheme, the answer module is further configured to extract a plurality of question-answer pairs from the knowledge information, where the question-answer pairs include candidate questions and corresponding candidate answers; storing the question-answer pairs into a question-answer library; and querying candidate questions similar to the questions in the question-answer library, and taking candidate answers corresponding to the queried candidate questions as answers corresponding to the questions.
In the above scheme, the number of the question-answer libraries is multiple, and the multiple question-answer libraries include a question-answer library corresponding to the live broadcast room and a question-answer library corresponding to the associated live broadcast room; the answer module is further used for inquiring candidate questions similar to the questions in a question-answer library corresponding to the live broadcast room, and taking candidate answers corresponding to the inquired candidate questions as answers of the questions; and when the candidate question similar to the question is not inquired in the question-answer library corresponding to the live broadcast room, inquiring the candidate question similar to the question in the question-answer library corresponding to the associated live broadcast room, and taking the candidate answer corresponding to the inquired candidate question as the answer of the question.
In the above scheme, the answer to the question is obtained by a machine learning model for reading understanding; the answer module is further configured to perform, by the machine learning model: and performing encoding processing on the basis of the knowledge information and the question to obtain an answer vector, and performing decoding processing on the answer vector to obtain an answer of the question.
In the above scheme, the answer module is further configured to perform encoding processing on the question to obtain a corresponding question vector; coding the knowledge information to obtain a corresponding knowledge vector; performing semantic analysis on the knowledge vector based on the question vector to obtain an answer vector of the question.
In the above solution, the answer module is further configured to determine a plurality of key vectors and a plurality of value vectors corresponding to the plurality of key vectors one to one from the knowledge vectors; selecting a plurality of adjacent key vectors from the plurality of key vectors, and combining the plurality of adjacent key vectors in a plurality of different ways to obtain a plurality of different key vector sequences; determining an attention value for each of the key vector sequences based on the problem vector and the plurality of value vectors; and selecting a target key vector sequence with the largest attention value from the plurality of key vector sequences, and combining a plurality of key vectors contained in the target key vector sequence to obtain the answer vector.
In the above solution, the answer module is further configured to perform the following processing for each key vector: determining a similarity between the problem vector and the key vector, and determining a product between the similarity and a corresponding value vector as an attention value of the key vector; performing the following for each of the key vector sequences: and summing the attention values of all key vectors contained in the key vector sequence, and determining the sum result as the attention value of the key vector sequence.
In the above solution, the answer module is further configured to determine that an operation of obtaining an answer to the question through the machine learning model is to be performed when no candidate question adapted to the question is queried in a question-and-answer library; wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
In the above scheme, the answer module is further configured to retrieve an answer corresponding to the question from a question-and-answer library, and present the retrieved answer; responding to a feedback operation aiming at the presented answer, and acquiring feedback information submitted by the feedback operation; when the feedback information is of a negative type, determining that an operation of obtaining an answer to the question through the machine learning model is to be performed; wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
In the above scheme, the answer module is further configured to retrieve an answer to the question from a question-and-answer library; determining that an operation of obtaining an answer to the question through the machine learning model is to be performed when the retrieved answer satisfies a failure condition; wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information; wherein the failure condition comprises at least one of: the time length of the retrieved answers existing in the question-answer library exceeds a time length threshold, the number of times that the retrieved answers are detected does not exceed a time threshold, and the number of the feedback information of the negative type corresponding to the retrieved answers exceeds a number threshold.
In the above solution, the information processing apparatus in live broadcasting further includes: the synchronization module is used for sending the problems to a client logged in with a main broadcasting account for presentation; responding to an answer operation aiming at the question, acquiring an answer submitted by the answer operation, and presenting the submitted answer; and forming question-answer pairs by the questions and the submitted answers, and synchronizing the question-answer pairs to a question-answer library.
In the above scheme, the synchronization module is further configured to respond to a feedback operation for the answer, and acquire feedback information submitted by the feedback operation; when the feedback information is of a negative type, sending the question to a client logged in with a main broadcasting account for presentation; responding to an answer operation aiming at the question, acquiring an answer submitted by the answer operation, and presenting the submitted answer; and forming question-answer pairs by the questions and the submitted answers, and synchronizing the question-answer pairs to a question-answer library.
An embodiment of the present application provides an electronic device, including:
a memory for storing computer executable instructions;
and the processor is used for realizing the information processing method in the live broadcast provided by the embodiment of the application when executing the computer executable instruction stored in the memory.
The embodiment of the application provides a computer-readable storage medium, which stores computer-executable instructions and is used for realizing the information processing method in live broadcast provided by the embodiment of the application when being executed by a processor.
The embodiment of the application has the following beneficial effects:
learning known identification information from the interactive events occurring in the live broadcast room, responding to the problems raised by the live broadcast room, and fully utilizing the characteristic that the interactive events can hit important information related to the live broadcast room, thereby improving the accuracy and efficiency of answers.
Drawings
Fig. 1 is an architecture diagram of an information processing system 100 in live broadcast provided by an embodiment of the present application;
fig. 2 is a schematic structural diagram of a terminal 400 provided in an embodiment of the present application;
fig. 3 is a schematic flowchart of an information processing method in live broadcast according to an embodiment of the present application;
fig. 4 is a schematic flowchart of an information processing method in live broadcast according to an embodiment of the present application;
fig. 5 is a schematic flowchart of an information processing method in live broadcast according to an embodiment of the present application;
fig. 6 is a schematic diagram illustrating an information processing method in live broadcast according to an embodiment of the present application;
fig. 7A, 7B and 7C are schematic diagrams illustrating an information processing method in live broadcast according to an embodiment of the present application;
fig. 8A, fig. 8B and fig. 8C are schematic diagrams illustrating an information processing method in live broadcast according to an embodiment of the present application;
fig. 9A, 9B, 9C, 9D, and 9E are application scene diagrams of an information processing method in live broadcast according to an embodiment of the present application.
Detailed Description
In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be described in further detail with reference to the attached drawings, the described embodiments should not be considered as limiting the present application, and all other embodiments obtained by a person of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.
In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the present application only and is not intended to be limiting of the application.
Before further detailed description of the embodiments of the present application, terms and expressions referred to in the embodiments of the present application will be described, and the terms and expressions referred to in the embodiments of the present application will be used for the following explanation.
1) In response to the condition or state on which the performed operation depends, one or more of the performed operations may be in real-time or may have a set delay when the dependent condition or state is satisfied; there is no restriction on the order of execution of the operations performed unless otherwise specified.
2) The terminal comprises a client, and an application program running in the terminal and used for providing various services, such as a live client or a video client.
3) Natural Language Processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that enable efficient communication between humans and computers using natural language. Natural language processing is a science integrating linguistics, computer science and mathematics. Therefore, the research in this field will involve natural language, i.e. the language that people use everyday, so it is closely related to the research of linguistics. Natural language processing techniques typically include text processing, semantic understanding, machine translation, robotic question and answer, knowledge mapping, and the like.
4) Machine Learning (ML) is a multi-domain cross discipline, and relates to a plurality of disciplines such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and the like. The special research on how a computer simulates or realizes the learning behavior of human beings so as to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve the performance of the computer. Machine learning is the core of artificial intelligence, is the fundamental approach for computers to have intelligence, and is applied to all fields of artificial intelligence. Machine learning and deep learning generally include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and formal education learning.
The fire heat of the live scene promotes a plurality of new services, such as live delivery or online education. The user can ask a question to the anchor or teacher through the input box, and the question asked by the user is basically replied by the manual work, which is time-consuming and labor-consuming.
The embodiment of the application provides an retrieval type Question-answering system, which finds a Question (Question) closest to a Question (Query) in a data set of a Question-Answer (QA, Question-Answer) pair (or called a Question-Answer pair) customized by an anchor or a teacher, and then recommends the Answer corresponding to the Question to a user.
The embodiment of the application can solve the problem that part of QA pairs which can be matched with the user-defined QA pairs. However, the coverage rate is low, and the new problem that the user is interested in the content in the live broadcast cannot be solved, so that the problem replying efficiency is low.
In view of the above technical problems, embodiments of the present application provide an information processing method in live broadcasting, which can improve efficiency and accuracy of answering questions in a live broadcasting process. An exemplary application of the information processing method in live broadcasting provided by the embodiment of the present application is described below, and the information processing method in live broadcasting provided by the embodiment of the present application can be implemented by various electronic devices, for example, can be applied to various types of user terminals (hereinafter also referred to as simply terminals) such as smart phones, tablet computers, in-vehicle terminals, smart wearable devices, notebook computers, and desktop computers. Next, an exemplary application when the electronic device is implemented as a terminal will be explained.
Referring to fig. 1, fig. 1 is a schematic architecture diagram of an information processing system 100 in live broadcast provided in an embodiment of the present application. The information processing system 100 in the live broadcast includes: the server 200, the network 300, and the terminal 400 will be separately described.
The server 200 is a background server of the client 410, and is configured to respond to a live content acquisition request of the client 410 and send live content of a live broadcast room to the client 410.
The network 300, which is used as a medium for communication between the server 200 and the terminal 400, may be a wide area network or a local area network, or a combination of both.
The terminal 400 is configured to operate a client 410, where the client 410 is a client with a live broadcast function, such as a live broadcast client or a video client. The client 410 is configured to receive live content sent by the server 200, and present live content in a live room in a live page; the system is also used for acquiring knowledge information from an interactive event occurring in a live broadcast room in the process of presenting live broadcast content; and the live broadcast server is also used for responding to the question submitting operation, presenting the submitted question in the live broadcast page, acquiring an answer of the question based on the knowledge information and presenting the answer.
In some embodiments, the terminal 400 implements the information processing method in live broadcast provided by the embodiments of the present application by running a computer program, for example, the computer program may be a native program or a software module in an operating system; can be a local (Native) Application program (APP), i.e. a program that needs to be installed in an operating system to run, such as a live APP or a video APP; or may be an applet, i.e. a program that can be run only by downloading it to the browser environment; but also a live applet or video applet that can be embedded into any APP. In general, the computer programs described above may be any form of application, module or plug-in.
The embodiments of the present application may be implemented by means of Cloud Technology (Cloud Technology), which refers to a hosting Technology for unifying series resources such as hardware, software, and network in a wide area network or a local area network to implement data calculation, storage, processing, and sharing.
The cloud technology is a general term of network technology, information technology, integration technology, management platform technology, application technology and the like applied based on a cloud computing business model, can form a resource pool, is used as required, and is flexible and convenient. Cloud computing technology will become an important support. Background services of the technical network system require a large amount of computing and storage resources.
As an example, the server 200 may be an independent physical server, may be a server cluster or a distributed system formed by a plurality of physical servers, and may also be a cloud server providing basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a web service, cloud communication, a middleware service, a domain name service, a security service, a CDN, and a big data and artificial intelligence platform. The terminal 400 and the server 200 may be directly or indirectly connected through wired or wireless communication, and the embodiment of the present application is not limited thereto.
The structure of the terminal 400 in fig. 1 is explained next. Referring to fig. 2, fig. 2 is a schematic structural diagram of a terminal 400 provided in an embodiment of the present application, where the terminal 400 shown in fig. 2 includes: at least one processor 410, memory 450, at least one network interface 420, and a user interface 430. The various components in the terminal 400 are coupled together by a bus system 440. It is understood that the bus system 440 is used to enable communications among the components. The bus system 440 includes a power bus, a control bus, and a status signal bus in addition to a data bus. For clarity of illustration, however, the various buses are labeled as bus system 440 in fig. 2.
The Processor 410 may be an integrated circuit chip having Signal processing capabilities, such as a general purpose Processor, a Digital Signal Processor (DSP), or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like, wherein the general purpose Processor may be a microprocessor or any conventional Processor, or the like.
The user interface 430 includes one or more output devices 431, including one or more speakers and/or one or more visual displays, that enable the presentation of media content. The user interface 430 also includes one or more input devices 432, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.
The memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid state memory, hard disk drives, optical disk drives, and the like. Memory 450 optionally includes one or more storage devices physically located remote from processor 410.
The memory 450 includes either volatile memory or nonvolatile memory, and may include both volatile and nonvolatile memory. The nonvolatile memory may be a Read Only Memory (ROM), and the volatile memory may be a Random Access Memory (RAM). The memory 450 described in embodiments herein is intended to comprise any suitable type of memory.
In some embodiments, memory 450 is capable of storing data, examples of which include programs, modules, and data structures, or a subset or superset thereof, to support various operations, as exemplified below.
The operating system 451, which includes system programs for handling various basic system services and performing hardware-related tasks, such as a framework layer, a core library layer, a driver layer, etc., is used for implementing various basic services and for processing hardware-based tasks.
A network communication module 452 for communicating to other computing devices via one or more (wired or wireless) network interfaces 420, exemplary network interfaces 420 including: bluetooth, wireless compatibility authentication (WiFi), and Universal Serial Bus (USB), among others.
A presentation module 453 for enabling presentation of information (e.g., user interfaces for operating peripherals and displaying content and information) via one or more output devices 431 (e.g., display screens, speakers, etc.) associated with user interface 430.
An input processing module 454 for detecting one or more user inputs or interactions from one of the one or more input devices 432 and translating the detected inputs or interactions.
In some embodiments, the information processing apparatus in the live broadcast provided by the embodiment of the present application may be implemented in software, and fig. 2 illustrates an information processing apparatus 455 in the live broadcast stored in the memory 450, which may be software in the form of programs and plug-ins, and includes the following software modules: a display module 4551, an acquisition module 4552 and a reply module 4553, which are logical and thus may be arbitrarily combined or further split depending on the functions implemented. The functions of the respective modules will be explained below.
Next, an example in which the terminal 400 in fig. 1 executes the information processing method in live broadcast provided in the embodiment of the present application is described. Referring to fig. 3, fig. 3 is a schematic flowchart of an information processing method in live broadcast according to an embodiment of the present application, and will be described with reference to the steps shown in fig. 3.
It should be noted that the method shown in fig. 3 can be executed by various forms of computer programs executed by the terminal 400, and is not limited to the client 410, such as the operating system 451, the software modules, and the scripts, described above, so that the following description taking the client as an example should not be taken as a limitation to the embodiments of the present application.
In step S101, live content of the live room is presented.
In some embodiments, a plurality of live rooms are included in the live platform, and live content of a selected live room is presented in response to a selection operation for the plurality of live rooms. In this way, the user can watch the live content which is interested by the user.
Taking the application scenario as live tape, the live content of the live room may be related content related to the anchor or anchor helper recommended items.
Taking the application scenario as online education, the live content of the live room may be related to the teaching of the teacher.
It should be noted that the anchor in the live broadcast scene may include a teacher giving lessons through the live broadcast room, and this embodiment of the present application does not distinguish between these.
In step S102, knowledge information is acquired from an interactive event occurring in the live broadcast room during presentation of live content.
In some embodiments, the client may invoke a corresponding service (e.g., a knowledge information acquisition service) of the terminal, and the acquisition process of the knowledge information is completed through the terminal. The client can also call a corresponding service (for example, a knowledge information acquisition service) of the server, and the acquisition process of the knowledge information is completed through the server.
In the following, a description will be given by taking an example in which a client invokes a corresponding service (for example, a knowledge information acquisition service) of a terminal and the acquisition process of knowledge information is completed by the terminal. It should be noted that the process of the client invoking the corresponding service (e.g., the knowledge information obtaining service) of the server to complete the obtaining of the knowledge information is similar to that described below, and will not be described again.
In some embodiments, knowledge information is obtained in real-time from interactive events occurring in the live room and/or associated live rooms during presentation of live content.
As an example, the interaction event comprises at least one of: and displaying the recommendation information, question and answer events and teaching events.
For example, when the interaction event is display recommendation information, the acquired knowledge information may include parameter information (e.g., price of a product, color of the product, shape of the product, and purchase manner of the product) of the anchor or an assistant of the anchor recommending the product in a live broadcast room, and may also be parameter information of the product on a background shelf. When the interactive event is a question-answer event, the question-answer event may occur between the anchor (or teacher) and the audience (or students) or between the audience (or students) and the audience (or students), for example, the anchor answers the audience's questions. When the interactive event is a teaching event, the acquired knowledge information may include information (for example, terms, definitions, principles, and the like in the course) of the teacher explaining the course in the live broadcast room, and may also be auxiliary teaching materials uploaded by the teacher explaining the course. Therefore, the diversity and the quantity of the acquired knowledge information can be improved, and the accuracy of the answers of the follow-up determined questions is improved.
As an example, associating the live room includes at least one of: the live broadcast rooms have competitive relationships with live broadcast rooms (e.g., live broadcast room where a key (PK) anchor is located or live broadcast room where recommended items are the same or similar), live broadcast rooms of anchor broadcasts having social relationships with the anchor broadcasts in the live broadcast rooms (e.g., live broadcast room where anchor friends are located), live broadcast rooms of the same type as the live broadcast rooms (e.g., live broadcast room of the same type as lessons explained by teachers, specifically, when the live broadcast rooms explain mathematics, the associated live broadcast room may be live broadcast room where the mathematics are explained as well), and live broadcast rooms of the same type as the anchor broadcasts (e.g., historical live broadcast room of the anchor broadcasts). Therefore, channels for acquiring knowledge information can be increased, diversity and quantity of acquired knowledge information are improved, and accuracy of answers of follow-up determined questions is improved.
In step S103, in response to the question submitting operation for the live broadcast, the question submitted by the question submitting operation is presented.
As an example, in fig. 9A, the user enters a question in an input box 901 and submits, and the submitted question 902 is presented in a live page. In this manner, the user can ask questions of the anchor while watching the live broadcast.
In step S104, an answer to the question is acquired based on the knowledge information, and the answer is presented.
In some embodiments, after the user submits the question and obtains the answer to the question, all answers may be presented in the unified answer area (as shown in fig. 9A, answer 903 is presented at the top of the interactive area); the corresponding answers may also be presented in adjacent locations to each question (as shown in fig. 9B, answers 903 are presented in adjacent locations to the questions).
In the embodiment of the application, the problems in the live broadcast room are automatically replied, and compared with the prior art in which the problems are replied manually, the method and the device can save human resources and improve the efficiency of problem replying; and the answer of the question is determined based on the knowledge information acquired from the interactive event occurring in the live broadcast room, so that the accuracy of the answer can be improved, and the live broadcast efficiency is improved.
In some embodiments, the client may invoke a corresponding service (e.g., an answer obtaining service) of the terminal, and the obtaining process of the answer is completed through the terminal. The client may also invoke a corresponding service (e.g., an answer obtaining service) of the server, and the obtaining process of the answer is completed through the server.
As an example, when the client calls a corresponding service (e.g., answer obtaining service) of the server to complete the obtaining process of the knowledge information, the alternative steps of step S104 are: sending the question to a server; the server acquires answers of the questions based on the knowledge information and sends the answers to the client; the client presents the answer.
In the following, a corresponding service (for example, an answer obtaining service) of the terminal is called by the client, and an obtaining process of an answer is completed by the terminal. It should be noted that the process of the client invoking the corresponding service (e.g., the answer obtaining service) of the server to obtain the answer is similar to that described below, and will not be described again.
In some embodiments, the answers to the questions are obtained through a question-answer library, specifically, a plurality of question-answer pairs are extracted from the knowledge information, wherein the question-answer pairs comprise candidate questions and corresponding candidate answers; storing a plurality of question-answer pairs to a question-answer library; and querying candidate questions similar to the questions in a question-answer library, and taking candidate answers corresponding to the queried candidate questions as answers of the corresponding questions.
As an example, a plurality of question-answer pairs are extracted from knowledge information in real time, and the plurality of question-answer pairs extracted in real time are stored in a question-answer library in real time, so that the question-answer library can be ensured to be updated in real time according to live broadcast content, and the question-answer pairs updated in real time are used for answering questions subsequently proposed by audiences in the live broadcast.
As an example, a similarity between the question and each candidate question in the question-and-answer library is determined, and the answer of the candidate question with the greatest similarity is selected as the answer of the question in the question-and-answer library.
For example, the FAQSet in fig. 7A is a question-and-answer library constructed in advance, Qi is a standard question in the question-and-answer library (i.e., the candidate question), and Ai is a corresponding standard answer (i.e., the candidate answer). Query new Is a problem posed by the user. Firstly, find out and Query from FAQSet new Most closely related Q i Then Q is added i The corresponding answer A i And returning to the user. The construction time of the question-answer library can be after the live broadcast starts and before the question of the user is obtained; or before the live broadcast begins.
Compared with the method for obtaining answers through a machine learning model, the method for obtaining answers of similar candidate questions from the question-answer library is high in speed, and therefore the obtaining speed of the answers can be improved, and the waiting time of a user is shortened.
As an example, when the number of the question-answer libraries is multiple and the multiple question-answer libraries include a question-answer library corresponding to the live broadcast room and a question-answer library corresponding to the associated live broadcast room, the candidate questions similar to the question are queried in the question-answer library, and taking the candidate answers corresponding to the queried candidate questions as the answers to the corresponding questions may be: inquiring candidate questions similar to the questions in a question-answer library corresponding to the live broadcast room, and taking candidate answers corresponding to the inquired candidate questions as answers of the questions; and when the candidate question similar to the question is not inquired in the question-answer library corresponding to the live broadcast room, inquiring the candidate question similar to the question in the question-answer library corresponding to the associated live broadcast room, and taking the candidate answer corresponding to the inquired candidate question as the answer of the question.
For example, the question-answer library corresponding to the live broadcast room comprises a plurality of question-answer pairs extracted from knowledge information acquired in real time from an interactive event occurring in the live broadcast room; the question-answer library corresponding to the associated live broadcast room comprises a plurality of question-answer pairs extracted from knowledge information acquired in real time from interactive events occurring in the associated live broadcast room.
According to the embodiment of the application, the answers are obtained from the question-answer library maintained on the basis of the knowledge information of the live broadcast room, and when the answers are not obtained, the answers are obtained from the question-answer library maintained on the basis of the knowledge information of the associated live broadcast room, so that the obtaining probability of the answers can be improved, and the accuracy of the answers can be improved.
In other embodiments, the answer to the question is obtained by a machine learning model for reading understanding, and in particular, the following is performed by the machine learning model: and performing encoding processing based on the knowledge information and the question to obtain an answer vector, and performing decoding processing on the answer vector to obtain an answer of the question.
As an example, the machine learning model for reading and understanding may be a question-and-Answer Network (QANet) model, and a specific implementation of obtaining answers to questions through the QANet model will be described in detail below.
As an example, performing encoding processing based on knowledge information and a question to obtain an answer vector may be performing encoding processing on the question to obtain a corresponding question vector; coding the knowledge information to obtain a corresponding knowledge vector; the knowledge vector is semantically analyzed based on the question vector to obtain an answer vector for the question.
As an example of accepting fig. 8A, the process of obtaining an answer to a question may be: coding the problem through a coding layer to obtain a corresponding problem vector; coding the knowledge information through a coding layer to obtain a corresponding knowledge vector; performing semantic analysis on the knowledge vector based on the question vector through an interaction layer to obtain an answer vector of the question; the answer vector is decoded by the output layer to obtain an answer to the question.
For example, semantically analyzing the knowledge vector based on the question vector to obtain an answer vector of the question may be determining a plurality of Key vectors and a plurality of Value vectors (Value) corresponding to the plurality of Key vectors one-to-one from the knowledge vector; selecting a plurality of adjacent key vectors from the plurality of key vectors, and combining the plurality of adjacent key vectors in a plurality of different ways to obtain a plurality of different key vector sequences; determining an attention value (or attention score) of each key vector sequence according to the problem vector and the value vectors; and selecting a target key vector sequence with the largest attention value from the plurality of key vector sequences, and combining a plurality of key vectors contained in the target key vector sequence to obtain an answer vector.
For example, from the problem vector and the plurality of value vectors, determining the attention value for each sequence of key vectors may be performing the following for each key vector: determining similarity between the problem vector and the key vector, and determining the product of the similarity and the corresponding value vector as the attention value of the key vector; the following processing is performed for each key vector sequence: the attention values of all key vectors included in the key vector sequence are summed, and the sum result is determined as the attention value of the key vector sequence.
For example, a key vector a, a key vector B, a key vector C, and a value vector a corresponding to the key vector a, a value vector B corresponding to the key vector B, a value vector C corresponding to the key vector C are determined from the knowledge vectors; the similarity a between the question vector and the key vector a, the similarity B between the question vector and the key vector B, and the similarity C between the question vector and the key vector C are determined, so that the attention value of the key vector a is aA, the attention value of the key vector B is bB, and the attention value of the key vector C is cC can be calculated. Selecting a plurality of adjacent key vectors from a plurality of key vectors for combination for multiple times to obtain key vector sequences (key vector A, key vector B), (key vector B, key vector C) and (key vector A, key vector B, key vector C); summing the attention values of all key vectors contained in the key vector sequence to obtain the attention value of the key vector sequence (key vector A, key vector B) as aA + bB; the attention value of the key vector sequence (key vector B, key vector C) is bB + cC; the attention value of the key vector sequence (key vector a, key vector B, key vector C) is aA + bB + cC; and selecting the key vector sequence with the maximum attention value, and combining a plurality of key vectors contained in the key vector sequence with the maximum attention value to obtain an answer vector.
Compared with the answer of similar candidate questions retrieved from the question-answer library, the accuracy of obtaining the answer through the machine learning model is higher, and therefore the accuracy of the answer can be improved, the satisfaction degree of the user on the answer is improved, and the live broadcast efficiency is improved.
As a first example, before obtaining answers to questions through the machine learning model, answers to the questions may also be retrieved in a question-and-answer library and the retrieved answers may be presented; responding to the feedback operation aiming at the presented answer, and acquiring feedback information submitted by the feedback operation; when the feedback information is of a negative type, determining that an operation of obtaining an answer to the question through a machine learning model is to be performed; wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
For example, the types of feedback information include a positive type (e.g., a question that is satisfied or that the answer has resolved a submission) and a negative type (e.g., a question that is not satisfied or that the answer has not resolved a submission).
For example, in fig. 9C, a feedback button 904 may be further displayed in the area where the answer 903 is displayed, and the feedback button 904 may be visible and operable for all viewers in the live room, or may be visible and operable only for viewers who submit questions corresponding to the answer 903 (e.g., viewer "wicks" who submitted the question "first put eggs or flour. When the audience triggers a 'satisfaction' button, the audience is represented to be satisfied with the answer 'flour first', namely the answer of the question is not required to be obtained through a machine learning model; when the audience triggers the 'dissatisfaction' button, the characteristic audience is dissatisfied with the answer 'first-time flour', namely the answer of the question needs to be obtained through a machine learning model.
In the embodiment of the application, compared with the answer of the similar candidate question retrieved from the question and answer library, the resource consumed by obtaining the answer through the machine learning model is more and the speed is lower, so that the answer is obtained through the machine learning model only when the answer retrieved and displayed from the question and answer library is not satisfied by the feedback of the user, the resource loss can be reduced, and the obtaining speed of the answer can be improved.
As a second example, before obtaining the answer to the question through the machine learning model, it may also be determined that the operation of obtaining the answer to the question through the machine learning model will be performed when no candidate question adapted to the question is queried in the question-and-answer library; wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
In the embodiment of the application, compared with the answer of the similar candidate question searched from the question and answer library, the resource consumed by obtaining the answer through the machine learning model is more and the speed is lower, so that the answer is obtained through the machine learning model only when the candidate question matched with the question is not inquired in the question and answer library, the resource loss can be reduced, and the obtaining probability of the answer can be improved.
As a third example, before obtaining answers to questions through the machine learning model, answers to questions may also be retrieved in a question-and-answer library; determining that an operation of obtaining an answer to the question through the machine learning model is to be performed when the retrieved answer satisfies the failure condition; wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
Here, the failure condition includes at least one of: the time length of the searched answers existing in the question-answering base exceeds a time length threshold value, the number of times that the searched answers are detected does not exceed a time threshold value, and the number of the feedback information of the negative type corresponding to the searched answers exceeds a number threshold value.
For example, the duration threshold may be a default value, a value set by a user, a client, or a server, or a value determined according to the duration of all candidate answers in the question-answering library, for example, an average value of the durations of all candidate answers is used as the duration threshold. The timeliness of the retrieved answers represented by the fact that the duration of the retrieved answers in the question and answer library exceeds the duration threshold is low, and the accuracy is low, namely the probability that the user is satisfied with the retrieved answers is low, so that the answers are abandoned, the answers to the questions are obtained through the machine learning model, the accuracy of the answers can be improved, and the satisfaction degree of the user can be improved.
For example, the number threshold may be a default value, a value set by a user, a client, or a server, or a value determined according to the number of times that all candidate answers in the question-answering library are detected, for example, an average value of the number of times that all candidate answers are detected is used as the number threshold. The number of times that the retrieved answers are detected in the question-answer library does not exceed the threshold value of the number of times represents that the retrieved answers are less used and have lower credibility, so that the answers are abandoned, the answers of the questions are obtained through the machine learning model, the accuracy of the answers can be improved, and the satisfaction degree of the user can be improved.
For example, the number threshold may be a default value, a value set by a user, a client, or a server, or a value determined according to the number of the negative types of feedback information corresponding to all candidate answers in the question-answering library, for example, an average value of the number of the negative types of feedback information corresponding to all candidate answers is used as the number threshold. The probability that the quantity of the negative type feedback information corresponding to the retrieved answer exceeds the quantity threshold value and represents that the user is satisfied with the retrieved answer is low, so that the answer is abandoned, the answer of the question is obtained through the machine learning model, the accuracy of the answer can be improved, and the satisfaction degree of the user can be improved.
In some embodiments, when an answer to a question is not obtained, the question may be sent to a client (or anchor client) logged in with an anchor account for presentation; responding to the answering operation aiming at the question, acquiring an answer submitted by the answering operation, and presenting the submitted answer; and forming question-answer pairs by the questions and the submitted answers, and synchronizing the question-answer pairs to a question-answer library.
By way of example, in fig. 9D, when an answer to the submitted question 902 is not obtained, the submitted question 902 is presented in the anchor client, along with a corresponding reply box 905, and when the anchor enters and submits the answer in the reply box 905, the answer 906 submitted by the anchor is presented in the live page. And synchronizes the submitted question 902 and the answer 906 submitted by the anchor to the question-answer library as question-answer pairs. Therefore, when the audience asks the same question again, the corresponding answer is directly obtained from the question-answer library, and repeated answers of the anchor are avoided.
In some embodiments, referring to fig. 4, fig. 4 is a flowchart illustrating an information processing method in live broadcast provided in an embodiment of the present application, and based on fig. 3, step S105 and step S106 may be included after step S104.
In step S105, in response to the feedback operation for the answer, feedback information submitted by the feedback operation is acquired.
As an example, in fig. 9E, a feedback button 904 may also be displayed in the area where the answer 903 is displayed, and the user may trigger a "happy" button and a "unhappy" button for the answer 903 to submit corresponding feedback information. For example, when the user triggers the "happy" button, the feedback information submitted is of the positive type; when the user triggers the "dissatisfied" button, the feedback information submitted is of a negative type.
In step S106, when the feedback information is of a negative type, an answer submitted by the solution operation is acquired, and the submitted answer is presented.
In some embodiments, when the feedback information is of a negative type, sending the question to a client logged in with the anchor account for presentation; responding to the answering operation aiming at the question, acquiring the answer submitted by the answering operation, and presenting the submitted answer; and forming question-answer pairs by the questions and the submitted answers, and synchronizing the question-answer pairs to a question-answer library.
As an example, in fig. 9E, when the user triggers the "dissatisfaction" button for answer 903, the submitted question 902 is presented in the anchor client, along with a corresponding reply box 905, and when the anchor enters and submits the answer in reply box 905, the answer 906 submitted by the anchor is presented in the live page. And synchronizes the submitted question 902 and the answer 906 submitted by the anchor to the question-answer library as question-answer pairs. Therefore, when the audience asks the same question again, the corresponding answer is directly obtained from the question-answer library, and repeated answers of the anchor are avoided.
The following describes an information processing method in live broadcast provided in an embodiment of the present application, taking an example that an application scene is live broadcast tape cargo.
According to the embodiment of the application, the intelligent small question-answering assistant is added in a live scene, some common problems of the user are answered, the anchor can be helped to reply the problems of the user (for example, live audiences), the reply of the anchor repeated problems is avoided, the live broadcast efficiency is improved, and the economic benefit is improved.
According to the embodiment of the application, the intelligent small question-answering assistant is realized through an NLP technology and a conversation technology, the anchor is assisted to answer some problems of the user, and meanwhile, the intelligent small question-answering assistant can also realize the functions of assisting the anchor to collect user problems, activating atmosphere (for example, when a live room comes in with a new audience, a welcome phrase of 'welcome XX comes to the live room' is sent), prompting new commodities on a background and the like. The core question-answering module of the intelligent question-answering small assistant can perform answer search in various question-answering data (such as an anchor customized question-answering pair and anchor live content) through various question-answering strategies (such as retrieval type question-answering and reading comprehension type question-answering), and can well adapt the intelligent question-answering to live scenes.
In some embodiments, the intelligent question-answering assistant can also collect new questions of the user, give answers by the host, and insert the new questions and answers into the question-answering library in real time through a voice translation technology, so that next time similar questions can be directly replied by the Answer (Answer) of the QA pair, and the problem that the host answers repeatedly can be avoided. Meanwhile, the embodiment of the application converts the voice content in the played live broadcast content into the text, and for some problems which cannot be solved by the question and answer library, answers are searched in the main broadcast voice text through the reading understanding technology in the NLP. Therefore, the intelligent small question-answer assistant which can update the question-answer library in real time and integrate the search-type question-answer system and the reading comprehension-type question-answer system can be added in the live broadcast field to help the anchor broadcast to solve the user problem, and the live broadcast efficiency is improved.
Referring to fig. 5, fig. 5 is a schematic flowchart of an information processing method in live broadcast provided in an embodiment of the present application, and will be described with reference to fig. 5.
In some embodiments, the user first puts forward a question through the input box, and then the intelligent question and answer assistant confirms that the input content is a question to be answered, rather than chatting content, through sentence analysis. When the input content is a question to be answered, the live broadcast system submits the question to a question-answering system to answer: firstly, the question-answering system answers through a search-type question-answering system, and the search-type question-answering system searches answers corresponding to the most similar questions in a question-answering library. When the answer can be found by the retrieval type question-answering system, the answer is directly returned to the user. When the retrieval type question-answering system can not give answers, the reading comprehension type question-answering system answers, and the reading comprehension type question-answering system mainly finds the answers from the explanation contents based on the explanation contents of the main broadcasting. If the answer is not found, the user question is sent to the anchor to remind the anchor of the help answer. After the anchor answers, the question-answer content enters the explanation content of the anchor again, and the question-answer pair is inserted into a question-answer library used by the retrieval type question-answer system.
Next, a specific implementation of the information processing method in live broadcast provided in the embodiment of the present application will be described.
(1) Sentence analysis (or question analysis).
The main purpose of sentence analysis is to determine whether the content input by the user is a Question asked by the user (i.e. a Question to be answered), rather than chatty. The embodiment of the application converts the problem of sentence analysis into the problem of sentence classification, for example, a FastText model is selected for sentence two classification.
Referring to fig. 6, fig. 6 is a schematic diagram illustrating a method for processing information in live broadcast according to an embodiment of the present application, and will be described with reference to fig. 6.
In some embodiments, the input is a plurality of words represented by vectors and their N-gram features, the output is whether the sentence is a question (e.g., output 0 indicates that the sentence is a question to be answered, output 1 indicates that the sentence is not a question to be answered), and the hidden layer performs the superposition averaging on the plurality of word vectors. And marking a plurality of positive and negative samples in the training data, obtaining a trained model through model training, and using the trained model for model prediction.
(2) A retrievable-Based QA.
The core of the search-based question answering is to find out the most relevant or closest answer to the question to the user according to the similarity between the question of the user and the questions in the question answering library.
Referring to fig. 7A, 7B and 7C, fig. 7A, 7B and 7C are schematic diagrams illustrating an information processing method in live broadcast according to an embodiment of the present application, and will be described with reference to fig. 7A, 7B and 7C.
In some embodiments, the FAQSet in fig. 7A is a question-and-answer library constructed in advance, Qi is a standard question in the question-and-answer library (i.e., the candidate question), and Ai is a corresponding standard answer (i.e., the candidate answer). Query new The embodiment of the application finds out the Query from the FAQSet n ew Most closely related Q i Finally, Q will be i Corresponding answer A i And returning to the user.
As an example, the construction timing of the question-answering library can be after the live broadcast starts and before the question of the user is obtained; or before the live broadcast begins.
In some embodiments, fig. 7B is a schematic diagram of a system architecture of a query-based answering system, and in fig. 7B, the system architecture of the query-based answering system is mainly divided into an offline service and an online service. The off-line service establishes an index base for the problems and an ordering model; after the online service obtains the user questions, a plurality of similar questions are obtained in the index database according to keywords in the questions, then the similar questions are reordered according to the ordering model, and the answers of the most similar questions are selected to be displayed to the user.
As an example, fig. 7C is a schematic structural diagram of a ranking model, the ranking model is modeled by using a similarity double-tower structure, the presentation layer uses an albert (a Lite bert) model, and the matching layer uses cosine similarity. Training data is manually labeled with a number of similar questions. The training method uses a loading training model and then training by Fine-tune (Fine-tune).
(3) Machine-readable rational question answering (MRC-Based QA).
In some embodiments, assuming that the question posed by the user appears in the narrative before the main broadcast, i.e., the answer to the question posed by the user is limited to a clause of the article, the model is required to indicate the correct answer start and end positions in the article. Here, the definition of the QA task may be: given a text C ═ C 1 ,c 2 ,…,c n Q ═ Q } and problem Q ═ Q 1 ,q 2 ,…,q m And finding a paragraph S ═ S from the text C i ,s i+1 ,…,s i+j And } answers question Q.
Referring to fig. 8A, 8B and 8C, fig. 8A, 8B and 8C are schematic diagrams illustrating an information processing method in live broadcast according to an embodiment of the present application, and will be described with reference to fig. 8A, 8B and 8C.
In some embodiments, fig. 8A is a system architecture diagram of a machine-readable comprehension type question-answering system. In fig. 8A, a question posed by a user and the entire article are coded and represented by a coding layer, where the coding layer is used to perform underlying processing on the article and the question, respectively, and convert a text into a digital code. The interaction layer enables the model to focus on semantic relation between articles and problems, understanding of the problems is deepened through semantic analysis of the articles, and understanding of the articles is deepened through the semantic analysis of the problems. And the output layer generates the answer output of the model according to the semantic analysis result and the type of the answer.
As an example, the machine-readable comprehension question-answering may be implemented by using a QANet model, and fig. 8B is a schematic structural diagram of the QA Net model. The QANet Model is composed of an Input Embedding Layer (Input Embedding Layer), an Embedding Encoder Layer (Embedding Encoder Layer), a question-content Attention Layer (Context-query entry Layer), a Model Encoder Layer (Model Encoder Layer), and an output Layer (Ou Input Layer), which will be described below.
1) Inputting an embedding layer:
in some embodiments, the pre-training words are embedded and output through a high speed Network (high Network), and the formula is expressed as x ═ high (x) w ,x c ) Wherein x is w Embedding a vector for a Word (Word embedded ding); x is the number of c The Character Embedding vector is the output of the Character level convolutional neural network (Char-CNN).
2) Embedding the encoder layer:
in some embodiments, the embedded Encoder layer includes a plurality of encoding modules (Encoder blocks), and fig. 8C is a schematic structural diagram of the encoding modules. The coding module comprises a plurality of Convolutional Neural Networks (CNN), a Self-attention (Self-attention) Layer and a feed-forward Layer (fed Layer), wherein the CNN is used for modeling a local text structure, the Self-attention Layer is used for modeling a global dependency relationship of the text, and each part adopts a residual error structure.
3) Question-content attention layer:
in some embodiments, a trilinear similarity function f (q, c) ═ W is first defined 0 [q,c,q⊙c]Constructing a similarity matrix S epsilon R between an article (Context) and a Query word n*m And obtaining a matrix through Row-wise Softmax normalization
Then calculating to obtain an attention moment array
In addition, a method in a Deep Cross Network (DCN) can be adopted to obtain the similarity matrix through Column-wise Softmax
The final result is a problem-content attention (Q2C attention) matrix
4) Model encoder layer:
in some embodiments, the model encoder layer takes [ c, a, c ^ B ] as input, where a and B are the line vectors from attention matrices A and B, respectively. Three model encodings (M odel encoders) in the model Encoder layer all employ a two-way attention mechanism and share model parameters.
5) An output layer:
in some embodiments, the Starting point probability value (Starting Position) is p 1 ＝softmax(W 1 [M 0 ；M 1 ]) (ii) a The endpoint probability value (Ending Position) is p 2 ＝softmax(W 2 [M 0 ；M 2 ]) Wherein M is 0 、M 1 、M 2 The outputs of the three Model encoders are provided.
In some embodiments, the loss function is
Wherein,
the meaning of this penalty function is to maximize the probability of giving a correct prediction of the head-to-tail position of the answer for each sample, for the start and end positions of the true answer. For prediction
And solving by adopting a dynamic programming algorithm.
According to the embodiment of the application, the answer library can be dynamically updated, for example, the intelligent question answering assistant collects questions which cannot be answered by the intelligent question answering assistant and gives the questions to the main broadcast to answer the questions. The live content of the anchor is translated into text, QA pairs in the live content are extracted in a template matching mode, and the new QA pairs are inserted into a question-answering library.
The question answering system is added into the live scene, the problems of the user can be conveniently and quickly solved, and therefore the live efficiency can be improved.
An exemplary structure of an information processing apparatus in live broadcast provided by an embodiment of the present application, implemented as a software module, is described below with reference to fig. 2.
In some embodiments, as shown in fig. 2, the software modules stored in the information processing device 455 in the live broadcast of memory 450 may include:
a display module 4551, configured to present live content in a live broadcast room; an obtaining module 4552, configured to obtain knowledge information from an interaction event occurring in a live broadcast room during presentation of live broadcast content; and an answer module 4553, configured to, in response to a question submitting operation for the live broadcast room, present a question submitted by the question submitting operation, acquire an answer to the question based on the knowledge information, and present the answer.
In the above scheme, the obtaining module 4552 is further configured to obtain knowledge information in real time from an interactive event occurring in the live broadcast room and/or the associated live broadcast room; wherein the interaction event comprises at least one of: displaying recommendation information, question and answer events and teaching events; wherein the associated live broadcast room comprises at least one of: the live broadcast room is in competition relation with the live broadcast rooms, the live broadcast room of the anchor in the live broadcast room is in social relation with the anchor, and the live broadcast room is of the same type as the live broadcast room.
In the above solution, the answer module 4553 is further configured to extract a plurality of question-answer pairs from the knowledge information, where a question-answer pair includes a candidate question and a corresponding candidate answer; storing a plurality of question-answer pairs to a question-answer library; and querying candidate questions similar to the questions in a question-answer library, and taking candidate answers corresponding to the queried candidate questions as answers of the corresponding questions.
In the scheme, the number of the question-answer libraries is multiple, and the multiple question-answer libraries comprise a question-answer library corresponding to the live broadcast room and a question-answer library corresponding to the associated live broadcast room; the answer module 4553 is further configured to query candidate questions similar to the question from a question-answer library corresponding to the live broadcast room, and use candidate answers corresponding to the queried candidate questions as answers to the question; and when the candidate question similar to the question is not inquired in the question-answer library corresponding to the live broadcast room, inquiring the candidate question similar to the question in the question-answer library corresponding to the associated live broadcast room, and taking the candidate answer corresponding to the inquired candidate question as the answer of the question.
In the above scheme, the answer to the question is obtained by a machine learning model for reading understanding; an answer module 4553, further configured to perform the following processing by the machine learning model: and performing encoding processing based on the knowledge information and the question to obtain an answer vector, and performing decoding processing on the answer vector to obtain an answer of the question.
In the above scheme, the answer module 4553 is further configured to perform encoding processing on the question to obtain a corresponding question vector; coding the knowledge information to obtain a corresponding knowledge vector; the knowledge vector is semantically analyzed based on the question vector to obtain an answer vector for the question.
In the above solution, the answer module 4553 is further configured to determine a plurality of key vectors and a plurality of value vectors corresponding to the plurality of key vectors one to one from the knowledge vectors; selecting a plurality of adjacent key vectors from the plurality of key vectors, and combining the plurality of adjacent key vectors in a plurality of different ways to obtain a plurality of different key vector sequences; determining an attention value of each key vector sequence according to the problem vector and the plurality of value vectors; and selecting a target key vector sequence with the largest attention value from the plurality of key vector sequences, and combining a plurality of key vectors contained in the target key vector sequence to obtain an answer vector.
In the above scheme, the answer module 4553 is further configured to perform the following processing for each key vector: determining similarity between the problem vector and the key vector, and determining the product of the similarity and the corresponding value vector as the attention value of the key vector; the following processing is performed for each key vector sequence: the attention values of all key vectors included in the key vector sequence are summed, and the sum result is determined as the attention value of the key vector sequence.
In the above solution, the answer module 4553 is further configured to determine that an operation of obtaining an answer to a question through a machine learning model is to be performed when a candidate question adapted to the question is not queried in the question-and-answer library; wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
In the above solution, the answer module 4553 is further configured to search the question-answer library for answers to the corresponding questions, and present the searched answers; responding to the feedback operation aiming at the presented answer, and acquiring feedback information submitted by the feedback operation; when the feedback information is of a negative type, determining that an operation of obtaining an answer to the question through a machine learning model is to be performed; wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
In the above scheme, the answer module 4553 is further configured to search the question-answer library for answers to the questions; determining that an operation of obtaining an answer to the question through the machine learning model is to be performed when the retrieved answer satisfies the failure condition; the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information; wherein the failure condition comprises at least one of: the time length of the searched answers existing in the question-answering base exceeds a time length threshold value, the number of times that the searched answers are detected does not exceed a time threshold value, and the number of the feedback information of the negative type corresponding to the searched answers exceeds a number threshold value.
In the above solution, the information processing apparatus 455 in live broadcasting further includes: the synchronization module is used for sending the problems to the client terminal logged in with the anchor account for presentation; responding to the answering operation aiming at the question, acquiring the answer submitted by the answering operation, and presenting the submitted answer; and forming question-answer pairs by the questions and the submitted answers, and synchronizing the question-answer pairs to a question-answer library.
In the above scheme, the synchronization module is further configured to respond to a feedback operation for the answer, and acquire feedback information submitted by the feedback operation; when the feedback information is of a negative type, sending the question to a client logged in with a main broadcasting account for presentation; responding to the answering operation aiming at the question, acquiring the answer submitted by the answering operation, and presenting the submitted answer; and forming question-answer pairs by the questions and the submitted answers, and synchronizing the question-answer pairs to a question-answer library.
Embodiments of the present application provide a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the information processing method in live broadcast in the embodiment of the present application.
The embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions, where the computer-executable instructions are stored, and when being executed by a processor, the computer-readable storage medium will cause the processor to execute the method for processing information in live broadcast provided by the embodiment of the present application, for example, the method for processing information in live broadcast shown in fig. 3, fig. 4, and fig. 5, and the computer includes various computing devices including an intelligent terminal and a server.
In some embodiments, the computer-readable storage medium may be memory such as FRAM, ROM, PROM, EP ROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or may be various devices including one or any combination of the above memories.
In some embodiments, the computer-executable instructions may be in the form of programs, software modules, scripts or code written in any form of programming language, including compiled or interpreted languages, or declarative or procedural languages, and they may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.
By way of example, computer-executable instructions may correspond, but do not necessarily have to correspond, to files in a file system, and may be stored in a portion of a file that holds other programs or data, e.g., in one or more scripts in a hypertext markup language document, in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub-programs, or portions of code).
By way of example, computer-executable instructions may be deployed to be executed on one computing device or on multiple computing devices at one site or distributed across multiple sites and interconnected by a communication network.
In summary, the embodiment of the present application has the following beneficial effects:
(1) the problem to the live broadcast room replies automatically, compares and replies through the manual work in correlation technique, not only can practice thrift manpower resources, can also improve the efficiency that the problem replied.
(2) The answer of the question is determined based on the knowledge information acquired from the interactive event occurring in the live broadcast room, so that the accuracy of the answer can be improved, and the live broadcast efficiency is improved.
(3) Compared with the method for acquiring answers through a machine learning model, the method for acquiring answers of similar candidate questions from the question-answer library has the advantages that the speed for retrieving answers of similar candidate questions is high, the acquiring speed of the answers can be improved, and therefore the waiting time of a user is shortened.
(4) The answers are obtained from the question-answer base maintained on the basis of the knowledge information of the live broadcast room, and when the answers are not obtained, the answers are obtained from the question-answer base maintained on the basis of the knowledge information of the associated live broadcast room, so that the obtaining probability of the answers can be improved, and the accuracy of the obtained answers can be improved.
(5) Compared with the answer of similar candidate questions retrieved from the question-answer library, the accuracy of the answer obtained through the machine learning model is higher, so that the accuracy of the answer can be improved, the satisfaction degree of the user on the answer is improved, and the live broadcast efficiency is improved.
The above description is only an example of the present application, and is not intended to limit the scope of the present application. Any modification, equivalent replacement, and improvement made within the spirit and scope of the present application are included in the protection scope of the present application.
Claims (15)
1. An information processing method in live broadcasting, the method comprising:
presenting live content of a live room;
acquiring knowledge information from an interactive event occurring in the live broadcast room in the process of presenting the live broadcast content;
in response to a question submission operation directed to the live broadcast room, presenting a question submitted by the question submission operation, and
and acquiring an answer of the question based on the knowledge information, and presenting the answer.
2. The method of claim 1, wherein the obtaining knowledge information from the interactive event occurring in the live broadcast room comprises:
acquiring knowledge information in real time from the interactive events occurring in the live broadcast room and/or the associated live broadcast room;
wherein the interaction event comprises at least one of: displaying recommendation information, question and answer events and teaching events;
wherein the associated live room comprises at least one of: the live broadcast room comprises a live broadcast room having a competitive relationship with the live broadcast room, a live broadcast room of a main broadcast having a social relationship with a main broadcast in the live broadcast room, and a live broadcast room of the same type as the live broadcast room.
3. The method of claim 1, wherein obtaining the answer to the question based on the knowledge information comprises:
extracting a plurality of question-answer pairs from the knowledge information, wherein the question-answer pairs comprise candidate questions and corresponding candidate answers;
storing the question-answer pairs into a question-answer library;
and querying candidate questions similar to the questions in the question-answer library, and taking candidate answers corresponding to the queried candidate questions as answers corresponding to the questions.
4. The method of claim 3,
the number of the question-answer libraries is multiple, and the multiple question-answer libraries comprise a question-answer library corresponding to the live broadcast room and a question-answer library corresponding to the associated live broadcast room;
the querying, in the question-answer library, candidate questions similar to the question, and taking candidate answers corresponding to the queried candidate questions as answers to the question, includes:
querying candidate questions similar to the questions in a question-answer library corresponding to the live broadcast room, and taking candidate answers corresponding to the queried candidate questions as answers of the questions;
and when the candidate question similar to the question is not inquired in the question-answer library corresponding to the live broadcast room, inquiring the candidate question similar to the question in the question-answer library corresponding to the associated live broadcast room, and taking the candidate answer corresponding to the inquired candidate question as the answer of the question.
5. The method of claim 1,
the answer to the question is obtained by a machine learning model for reading comprehension;
the obtaining an answer to the question based on the knowledge information includes:
performing, by the machine learning model: and performing encoding processing on the basis of the knowledge information and the question to obtain an answer vector, and performing decoding processing on the answer vector to obtain an answer of the question.
6. The method of claim 5, wherein the encoding based on the knowledge information and the question to obtain an answer vector comprises:
coding the problem to obtain a corresponding problem vector;
coding the knowledge information to obtain a corresponding knowledge vector;
performing semantic analysis on the knowledge vector based on the question vector to obtain an answer vector of the question.
7. The method of claim 6, wherein the semantically analyzing the knowledge vector based on the question vector to obtain an answer vector for the question, comprises:
determining a plurality of key vectors and a plurality of value vectors corresponding to the plurality of key vectors one to one from the knowledge vectors;
selecting a plurality of adjacent key vectors from the plurality of key vectors, and combining the plurality of adjacent key vectors in a plurality of different ways to obtain a plurality of different key vector sequences;
determining an attention value for each of the key vector sequences based on the problem vector and the plurality of value vectors;
and selecting a target key vector sequence with the largest attention value from the plurality of key vector sequences, and combining a plurality of key vectors contained in the target key vector sequence to obtain the answer vector.
8. The method of claim 7, wherein determining the attention value for each of the sequences of key vectors based on the question vector and the plurality of value vectors comprises:
performing the following for each of the key vectors: determining a similarity between the problem vector and the key vector, and determining a product between the similarity and a corresponding value vector as an attention value of the key vector;
performing the following for each of the key vector sequences: and summing the attention values of all key vectors contained in the key vector sequence, and determining the sum result as the attention value of the key vector sequence.
9. The method of claim 5, wherein prior to obtaining the answer to the question through the machine learning model, the method further comprises:
determining that an operation of obtaining an answer to the question through the machine learning model is to be performed when a candidate question adapted to the question is not queried in a question-and-answer library;
wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
10. The method of claim 5, wherein prior to obtaining an answer to the question via the machine learning model, the method further comprises:
searching answers corresponding to the questions in a question-answer library, and presenting the searched answers;
responding to a feedback operation aiming at the presented answer, and acquiring feedback information submitted by the feedback operation;
when the feedback information is of a negative type, determining that an operation of obtaining an answer to the question through the machine learning model is to be performed;
wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information.
11. The method of claim 5, wherein prior to obtaining the answer to the question through the machine learning model, the method further comprises:
retrieving answers to said questions in a question-and-answer library;
determining that an operation of obtaining an answer to the question through the machine learning model is to be performed when the retrieved answer satisfies a failure condition;
wherein the question-answer library comprises a plurality of question-answer pairs extracted from the knowledge information;
wherein the failure condition comprises at least one of: the time length of the retrieved answers existing in the question-answer library exceeds a time length threshold, the number of times that the retrieved answers are detected does not exceed a time threshold, and the number of the feedback information of the negative type corresponding to the retrieved answers exceeds a number threshold.
12. The method of claim 1, wherein when an answer to the question is not obtained, the method further comprises:
sending the problem to a client logged in with a main broadcasting account for presentation;
responding to an answer operation aiming at the question, acquiring an answer submitted by the answer operation, and presenting the submitted answer;
and forming question-answer pairs by the questions and the submitted answers, and synchronizing the question-answer pairs to a question-answer library.
13. An information processing apparatus in live broadcasting, comprising:
the display module is used for presenting the live broadcast content of the live broadcast room;
the acquisition module is used for acquiring knowledge information from an interactive event occurring in the live broadcast room in the process of presenting the live broadcast content;
and the answer module is used for responding to the question submitting operation aiming at the live broadcast room, presenting the question submitted by the question submitting operation, acquiring the answer of the question based on the knowledge information and presenting the answer.
14. An electronic device, comprising:
a memory for storing computer executable instructions;
a processor for implementing the method of processing information in live broadcast of any of claims 1 to 12 when executing computer executable instructions stored in the memory.
15. A computer-readable storage medium having stored thereon computer-executable instructions for implementing the method of processing information in a live broadcast of any one of claims 1 to 12 when executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110127263.9A CN114817488A (en) | 2021-01-29 | 2021-01-29 | Information processing method and device in live broadcast, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110127263.9A CN114817488A (en) | 2021-01-29 | 2021-01-29 | Information processing method and device in live broadcast, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114817488A true CN114817488A (en) | 2022-07-29 |
Family
ID=82525610
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110127263.9A Pending CN114817488A (en) | 2021-01-29 | 2021-01-29 | Information processing method and device in live broadcast, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114817488A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115412745A (en) * | 2022-08-12 | 2022-11-29 | 联想(北京)有限公司 | Information processing method and electronic equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200014046A (en) * | 2018-07-31 | 2020-02-10 | 주식회사 포티투마루 | Device and Method for Machine Reading Comprehension Question and Answer |
| CN111611355A (en) * | 2019-02-25 | 2020-09-01 | 北京嘀嘀无限科技发展有限公司 | Dialog reply method, device, server and storage medium |
| CN111611361A (en) * | 2020-04-01 | 2020-09-01 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Intelligent reading, understanding, question answering system of extraction type machine |
| CN111680133A (en) * | 2019-03-11 | 2020-09-18 | 阿里巴巴集团控股有限公司 | Live broadcast question and answer method and device |
| CN111858859A (en) * | 2019-04-01 | 2020-10-30 | 北京百度网讯科技有限公司 | Automatic question-answering processing method, device, computer equipment and storage medium |
| CN112073741A (en) * | 2020-08-31 | 2020-12-11 | 腾讯科技（深圳）有限公司 | Live broadcast information processing method and device, electronic equipment and storage medium |
-
2021
- 2021-01-29 CN CN202110127263.9A patent/CN114817488A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200014046A (en) * | 2018-07-31 | 2020-02-10 | 주식회사 포티투마루 | Device and Method for Machine Reading Comprehension Question and Answer |
| CN111611355A (en) * | 2019-02-25 | 2020-09-01 | 北京嘀嘀无限科技发展有限公司 | Dialog reply method, device, server and storage medium |
| CN111680133A (en) * | 2019-03-11 | 2020-09-18 | 阿里巴巴集团控股有限公司 | Live broadcast question and answer method and device |
| CN111858859A (en) * | 2019-04-01 | 2020-10-30 | 北京百度网讯科技有限公司 | Automatic question-answering processing method, device, computer equipment and storage medium |
| CN111611361A (en) * | 2020-04-01 | 2020-09-01 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Intelligent reading, understanding, question answering system of extraction type machine |
| CN112073741A (en) * | 2020-08-31 | 2020-12-11 | 腾讯科技（深圳）有限公司 | Live broadcast information processing method and device, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 顾迎捷;桂小林;李德福;沈毅;廖东;: "基于神经网络的机器阅读理解综述", 软件学报, no. 07, 15 July 2020 (2020-07-15) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115412745A (en) * | 2022-08-12 | 2022-11-29 | 联想(北京)有限公司 | Information processing method and electronic equipment |
| CN115412745B (en) * | 2022-08-12 | 2024-02-27 | 联想(北京)有限公司 | Information processing method and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112073741B (en) | Live broadcast information processing method and device, electronic equipment and storage medium | |
| US10249207B2 (en) | Educational teaching system and method utilizing interactive avatars with learning manager and authoring manager functions | |
| CN112163165B (en) | Information recommendation method, device, equipment and computer readable storage medium | |
| US10360809B2 (en) | Course skeleton for adaptive learning | |
| Shih et al. | An investigation of attitudes of students and teachers about participating in a context‐aware ubiquitous learning activity | |
| CN113010702B (en) | Interactive processing method and device for multimedia information, electronic equipment and storage medium | |
| US20190303768A1 (en) | Community Question Answering-Based Article Recommendation Method, System, and User Device | |
| CN109947915B (en) | Knowledge management system-based artificial intelligence expert system and construction method thereof | |
| CN110362671B (en) | Topic recommendation method, device and storage medium | |
| CN107562836B (en) | Respondent recommendation method based on topic model and machine learning | |
| CN114281957A (en) | Natural language data query method and device, electronic equipment and storage medium | |
| CN111279404B (en) | Language fluent system | |
| CN110929045B (en) | Construction method and system of poetry-semantic knowledge map | |
| CN110795548A (en) | Intelligent question answering method, device and computer readable storage medium | |
| CN114430490B (en) | Live question-answering and interface display method and computer storage medium | |
| CN116541493A (en) | Interactive response method, device, equipment and storage medium based on intention recognition | |
| Hosseini et al. | A study of concept-based similarity approaches for recommending program examples | |
| CN112132075B (en) | Method and medium for processing image-text content | |
| CN114817488A (en) | Information processing method and device in live broadcast, electronic equipment and storage medium | |
| CN112307166B (en) | Intelligent question-answering method and device, storage medium and computer equipment | |
| Cole | Activity of understanding a problem during interaction with an “enabling” information retrieval system: Modeling information flow | |
| CN116610784A (en) | Insurance business scene question-answer recommendation method and related equipment thereof | |
| CN116701593A (en) | Chinese question-answering model training method based on GraphQL and related equipment thereof | |
| CN115640403A (en) | Knowledge management and control method and device based on knowledge graph | |
| CN113411618B (en) | Data processing method and device based on social application and computer storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 40070424 Country of ref document: HK |