WO2022126962A1 - 基于知识图谱的引导教唆语料的检测方法及其相关设备 - Google Patents
基于知识图谱的引导教唆语料的检测方法及其相关设备 Download PDFInfo
- Publication number
- WO2022126962A1 WO2022126962A1 PCT/CN2021/090164 CN2021090164W WO2022126962A1 WO 2022126962 A1 WO2022126962 A1 WO 2022126962A1 CN 2021090164 W CN2021090164 W CN 2021090164W WO 2022126962 A1 WO2022126962 A1 WO 2022126962A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- corpus
- detected
- guiding
- entity
- knowledge graph
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000001514 detection method Methods 0.000 claims abstract description 32
- 238000000605 extraction Methods 0.000 claims abstract description 11
- 230000015654 memory Effects 0.000 claims description 23
- 238000012790 confirmation Methods 0.000 claims description 20
- 230000011218 segmentation Effects 0.000 claims description 14
- 230000008094 contradictory effect Effects 0.000 claims description 7
- 239000000463 material Substances 0.000 claims description 2
- 239000003795 chemical substances by application Substances 0.000 description 20
- 230000006399 behavior Effects 0.000 description 7
- 238000004891 communication Methods 0.000 description 5
- 238000007689 inspection Methods 0.000 description 5
- 238000012795 verification Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000002708 enhancing effect Effects 0.000 description 4
- 238000003058 natural language processing Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 210000000988 bone and bone Anatomy 0.000 description 2
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 208000003028 Stuttering Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 230000003068 static 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
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
Definitions
- the present application relates to the field of big data technology, and in particular, to a method for detecting guidance and abetting corpus based on knowledge graphs and related equipment.
- the purpose of the embodiments of the present application is to propose a knowledge graph-based detection method for guiding and abetting corpus and related equipment, so as to quickly determine whether the corpus to be detected belongs to the corpus of guiding and instigating, and effectively realize the detection of guiding and instigating behavior.
- the embodiment of the present application provides a method for detecting guidance and abetting corpus based on knowledge graph, and adopts the following technical solutions:
- a detection method for guiding and abetting corpus based on knowledge graph comprising the following steps:
- Receive a standard corpus data set perform feature extraction on the standard corpus data set, and obtain standard corpus features, wherein there is no guidance and instigation information in the standard corpus data set;
- the deduction result is a deduction failure
- use the failed deduction entity to be detected as the guiding and abetting entity, and use the to-be-detected corpus corresponding to the guiding and instigating entity as the guiding and instigating corpus, and output the guiding and instigating corpus;
- the first knowledge graph is updated based on the successfully deduced to-be-detected entity, and the second knowledge graph is obtained.
- the embodiment of the present application also provides a detection device for guiding and abetting corpus based on knowledge graph, which adopts the following technical solutions:
- a detection device for guiding and abetting corpus based on knowledge graph comprising:
- a receiving module configured to receive a standard corpus data set, perform feature extraction on the standard corpus data set, and obtain standard corpus features, wherein there is no guiding and abetting information in the standard corpus data set;
- a building module for constructing a first knowledge graph based on the standard corpus features
- the recognition model is used to receive the corpus to be detected, to perform named entity recognition on the corpus to be detected, to obtain the entity to be detected, and to deduce each entity to be detected in the first knowledge graph to obtain the deduction result ;
- the output module is used for, when the deduction result is that the deduction fails, take the entity to be detected that the deduction failed as the guiding and abetting entity, and use the to-be-detected corpus corresponding to the guiding and instigating entity as the guiding and instigating corpus, and output the guiding and instigating corpus. instigation material;
- the updating module is configured to update the first knowledge graph based on the to-be-detected entity that is successfully deduced to obtain a second knowledge graph when the deduction result is a successful deduction.
- the embodiment of the present application also provides a computer device, which adopts the following technical solutions:
- a computer device comprising a memory and a processor, wherein computer-readable instructions are stored in the memory, and when the processor executes the computer-readable instructions, the method for detecting a knowledge graph-based guidance and abetting corpus as described below is implemented. step:
- Receive a standard corpus data set perform feature extraction on the standard corpus data set, and obtain standard corpus features, wherein there is no guidance and instigation information in the standard corpus data set;
- the deduction result is a deduction failure
- use the failed deduction entity to be detected as the guiding and abetting entity, and use the to-be-detected corpus corresponding to the guiding and instigating entity as the guiding and instigating corpus, and output the guiding and instigating corpus;
- the first knowledge graph is updated based on the successfully deduced to-be-detected entity, and the second knowledge graph is obtained.
- the embodiments of the present application also provide a computer-readable storage medium, which adopts the following technical solutions:
- Receive a standard corpus data set perform feature extraction on the standard corpus data set, and obtain standard corpus features, wherein there is no guidance and instigation information in the standard corpus data set;
- the deduction result is a deduction failure
- use the failed deduction entity to be detected as the guiding and abetting entity, and use the to-be-detected corpus corresponding to the guiding and instigating entity as the guiding and instigating corpus, and output the guiding and instigating corpus;
- the first knowledge graph is updated based on the successfully deduced to-be-detected entity, and the second knowledge graph is obtained.
- the present application proposes to detect the corpus to be detected based on the first knowledge graph, so as to determine whether the corpus to be detected belongs to the guidance and instigation corpus. Effectively realize the detection of the guidance and abetting behavior of agents in practical applications.
- by deriving the first knowledge map of the successful entity to be detected it is conducive to the continuous updating and learning of the model with the changes of the times, thereby enhancing the deterrence of the agents, further reducing the customer complaint rate, and effectively restricting the normative terms of the agents. Thereby increasing customer satisfaction.
- FIG. 1 is an exemplary system architecture diagram to which the present application can be applied;
- FIG. 2 is a flowchart of an embodiment of a method for detecting a knowledge graph-based guidance and abetting corpus according to the present application
- FIG. 3 is a schematic structural diagram of an embodiment of a detection device for guiding and abetting corpus based on a knowledge graph according to the present application;
- FIG. 4 is a schematic structural diagram of an embodiment of a computer device according to the present application.
- the system architecture 100 may include terminal devices 101 , 102 , and 103 , a network 104 and a server 105 .
- the network 104 is a medium used to provide a communication link between the terminal devices 101 , 102 , 103 and the server 105 .
- the network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
- the user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages and the like.
- Various communication client applications may be installed on the terminal devices 101 , 102 and 103 , such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, and the like.
- the terminal devices 101, 102, and 103 may be various electronic devices that have a display screen and support web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III, dynamic Picture Experts Compression Standard Audio Layer 3), MP4 (Moving Picture Experts Group Audio Layer IV, Moving Picture Experts Compression Standard Audio Layer 4) Players, Laptops and Desktops, etc.
- MP3 players Moving Picture Experts Group Audio Layer III, dynamic Picture Experts Compression Standard Audio Layer 3
- MP4 Moving Picture Experts Group Audio Layer IV, Moving Picture Experts Compression Standard Audio Layer 4
- the server 105 may be a server that provides various services, such as a background server that provides support for the pages displayed on the terminal devices 101 , 102 , and 103 .
- the method for detecting the knowledge graph-based guidance and abetting corpus is generally performed by a server/terminal device, and accordingly, the knowledge graph-based guidance and abetting corpus detection device is generally set on the server/terminal device. middle.
- terminal devices, networks and servers in FIG. 1 are merely illustrative. There can be any number of terminal devices, networks and servers according to implementation needs.
- FIG. 2 there is shown a flow chart of an embodiment of a method for detecting guidance and abetting corpus based on a knowledge graph according to the present application.
- the described detection method based on the knowledge graph-based guiding and abetting corpus comprises the following steps:
- S1 Receive a standard corpus data set, perform feature extraction on the standard corpus data set, and obtain standard corpus features, wherein there is no guidance and abetting information in the standard corpus data set.
- the labeled corpus data set in this application refers to a corpus data set without guidance and abetting information, that is, a compliant corpus.
- the electronic device for example, the server/terminal device shown in FIG. 1
- the knowledge graph-based detection method for guiding and abetting corpus runs can receive the labeled corpus data set through a wired connection or a wireless connection.
- the above wireless connection methods may include but are not limited to 3G/4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or developed in the future .
- the steps of extracting the standard corpus data set to obtain the standard corpus features include:
- the triple data of each corpus in the standard corpus data set is extracted as the standard corpus feature.
- the SPO Subject-Predicate-Object, Subject-Predicate-Object
- the SPO Subject-Predicate-Object
- Group data generate a triplet data set based on multiple triplet data, as a standard corpus feature.
- This application uses triple data as a standard corpus feature, which is convenient for the subsequent construction of the first knowledge graph.
- the step of extracting the triple data of each corpus in the standard corpus data set as the feature of the standard corpus includes:
- the triplet data is screened based on a preset limited relationship to obtain target triplet data, and the target triplet data is used as the standard corpus feature.
- the entity recognition tool in this application refers to jiagu (oracle bone).
- Jiagu (Oracle Bone) is a deep learning natural language processing tool, which also has the functions of Chinese word segmentation, part-of-speech tagging and named entity recognition.
- Jiagu is based on the BiLSTM (Bi-directional Long Short-Term Memory) model and is trained using large-scale corpus.
- Use jiagu to perform word segmentation on the standard corpus data set to obtain standard corpus words.
- jiagu to perform named entity recognition on the standard corpus words to obtain a named entity set.
- An example of the word segmentation operation is as follows: The original corpus is that Zhang Xian is a stylish Chinese.
- the word segmentation operation After the word segmentation operation, it becomes ['Zhang Xian', 'Yes', 'A', 'Cute', 'De', 'Chinese' ].
- the named entity set is obtained [Zhang Xian, Chinese].
- Determine the connection relationship between different named entities For example, the connection keyword between the named entities "Zhang Xian” and “Chinese” is "Yes”, then the connection relationship belongs to the subordinate relationship, and the triple data is Zhang Xian- Yes - Chinese.
- the named entities in the named entity set that conform to the limited relationship are connected to obtain triple data.
- the defined relationships in this application may include common-sense relationships such as parent-child relationship, affiliation, and the like.
- triplet data is as follows: Xidian University-Coordinates-Xi'an; Xidian University-School Type-985 Project; Zhang Moumou-Educational-Graduate. Since there is no guiding and instigating data in the standard corpus data set, the generated standard corpus features belong to the non-guiding and instigating features.
- the present application can also use the jieba (stuttering) word segmentation tool according to actual needs, which can be applied.
- a first knowledge graph is constructed based on standard corpus features, and the first knowledge graph is a knowledge graph of compliant discourse.
- the specific steps include: overlapping the same subject and/or object between different SPO triples.
- the specific coincidence methods can be subject-subject coincidence, subject-object coincidence, and object-object coincidence.
- the step of constructing the first knowledge graph based on the standard corpus features includes:
- the first knowledge graph is constructed based on a preset graph database and the standard corpus features.
- the graph database of the present application is the Neo4j library
- the graph created by the Neo4j library is a directed graph constructed with vertices and edges.
- a first knowledge graph is constructed by using the Neo4j library and the above-mentioned standard corpus features (that is, extracted into triples), and the first knowledge graph is a knowledge graph that does not involve guiding and abetting data.
- the first knowledge map established through the Neo4j library can facilitate subsequent update and expansion.
- the application generates an expandable knowledge map, which is conducive to the continuous update and learning of computers with the changes of the times.
- S3 Receive the corpus to be detected, perform named entity recognition on the corpus to be detected, obtain the entity to be detected, and deduce each entity to be detected in the first knowledge graph to obtain a deduction result.
- a task to be inspected is received, and the task to be inspected includes the corpus to be inspected.
- Use jiagu library to perform word segmentation and named entity recognition on the to-be-detected corpus to obtain a to-be-detected entity set, and traverse each to-be-detected entity in the to-be-detected entity set through the first knowledge graph to identify the to-be-detected entity set Detect whether an entity can be deduced in the knowledge graph.
- the specific deduction process is: finding the path of the entity to be detected in the first knowledge graph.
- the first knowledge graph contains such a path "Melinda Gates-Spouse-Bill Gates-Chairman-Microsoft-Headquarters-Seattle", and the entity to be detected is Melinda Gates.
- Melinda Gates lives in Seattle, and the deduction result is output as a successful deduction.
- the entity to be detected is the president, it is determined by searching in the first knowledge graph that there is no such entity.
- the similarity algorithm and determine the target entity whose semantic similarity exceeds the preset threshold by calculating the semantic similarity between each target entity in the first knowledge graph and the entity to be detected "item", as a substitute entity, find The path of the entity is replaced in the first knowledge graph, so that the deduction result is determined to be a successful deduction. If there is no target entity whose semantic similarity with the entity "item" to be detected exceeds the preset threshold in the first knowledge graph, the output deduction result is deduction failure.
- this application includes but is not limited to the above deduction process.
- any deduction method can be selected according to actual needs, and it can be applied.
- the corresponding to-be-detected corpus is determined through the to-be-detected entity that fails to deduce, so as to determine the guidance and instigation corpus. Realize the rapid identification of the guidance and instigation corpus. Thereby effectively restricting the normative language of the agents, reducing the customer complaint rate and improving customer satisfaction.
- the present application may use the scene corresponding to the guidance and abetment corpus as the guidance and abetment scene.
- the step of outputting the guiding and abetting corpus include:
- the to-be-detected entity that fails to be deduced is used as the guiding and instigating entity, and the to-be-detected knowledge graph is generated based on the to-be-detected corpus corresponding to the guiding and instigating entity;
- the corpus to be detected corresponding to the guiding and abetting entity is used as the corpus of guiding and abetting.
- triplet data in the corpus to be detected is extracted as triplet data to be detected.
- the knowledge graph to be detected is constructed based on the triplet data to be detected.
- the entity to be detected that cannot be deduced on the first knowledge graph is the guiding and abetting entity, and then according to the spatial positional relationship between entities, it is judged whether the corpus to be detected is the corpus of guiding and abetting, specifically, whether there is a contradiction between the to-be-detected knowledge graph and the first knowledge graph. . If there is a contradictory relationship, it is determined that the corpus to be detected is the corpus of guidance and abetment.
- the scene corresponding to the to-be-detected corpus is a guiding and abetting scene.
- the to-be-detected corpus is used as the to-be-confirmed corpus, and is saved in a preset database.
- the deduction can be deduced forward, that is, deduced from the subject to the object, or reversed, that is, deduced from the object to the subject.
- the computer has determined the part-of-speech of each word in the process of word segmentation through the jiagu library, and marked the part-of-speech of each word, that is to say whether each word belongs to the subject, object, predicate or Adjectives, etc.
- the contradictory relationship in this application refers to the logical expression conflict relationship between different knowledge graphs. For example, there is triple data of "Zhang Xian's education-yes-primary school" in the knowledge graph to be detected, and In the first knowledge graph, there is triple data of "Zhang Xian's education-yes-graduate student". At this time, the triple data in different knowledge graphs are contradictory, and then it is determined that the knowledge graph to be detected is the same as that of the first knowledge graph. There is a contradiction between knowledge graphs.
- the first knowledge graph is expanded by updating the successfully deduced entities to be detected, so as to realize the continuous updating of the knowledge graph, and further realize the self-learning updating and optimization of the guiding and abetting corpus by the computer.
- the first knowledge graph is updated based on the successfully deduced entity to be detected, and the step of obtaining the second knowledge graph includes:
- the corpus to be detected corresponding to the entity to be detected that has been successfully deduced is identified as the initial qualified corpus
- the initial qualified corpus is used as the target qualified corpus
- the first knowledge graph is updated based on the target qualified corpus to obtain a second knowledge graph.
- the first knowledge graph is updated based on the target qualified corpus
- the specific steps of obtaining the second knowledge graph include: converting the target qualified corpus into triple data, and adding the triple data to the first knowledge In the graph, the second knowledge graph is obtained.
- the initial qualified corpus can be quickly determined.
- the initial qualified corpus can be directly used as The target qualified corpus can quickly determine the target qualified corpus.
- step S4 the to-be-detected corpus corresponding to the guiding and abetting entity is taken as the guiding and instigating corpus, and after outputting the guiding and instigating corpus, the electronic device may also perform the following: step:
- the guiding and instigating corpus is a real guiding and instructing corpus, and when the guiding and instigating corpus is an unreal guiding and instigating corpus, the guiding and instigating corpus is added to the first knowledge graph to obtain an expanded knowledge graph.
- the determined guidance and abetting corpus after verification is not the real guidance and abetting corpus, it is considered that the guidance and abetting corpus is actually a compliant corpus, and the corpus is added to the first knowledge graph to realize the first An expansion of the knowledge graph. That is, review the quality inspection result of the above-mentioned judgment corpus belonging to the guidance and instigation corpus, and add the knowledge to the first knowledge graph for scenarios that do not violate the rules.
- the step of verifying whether the guiding and abetting corpus is a real guiding and abetting corpus includes:
- the second detection and verification of the corpus of guidance and abetting is performed through a pre-trained detection model of the corpus of guidance and abetment. If the result outputted by the instigation corpus detection model at this time is that the guiding and instigating corpus is the real guiding and instigating corpus, it can be further determined that the corpus belongs to the guiding and instigating corpus. Furthermore, it can be further determined that the scene corresponding to the corpus of the guidance and instigation type belongs to the scene of the guidance and instigation type.
- the instigation corpus detection model of the present application is an NLP (Natural Language Processing, natural language processing) model.
- the step of verifying whether the guiding and abetting corpus is a real guiding and abetting corpus includes:
- a confirmation signal sent by the user terminal When a confirmation signal sent by the user terminal is received, it is determined based on the confirmation signal whether the guiding and instigating corpus is a real guiding and instructing corpus, wherein the confirmation signal corresponds to the signal for requesting confirmation of the instructing corpus.
- the guidance and abetting corpus is output to the display device of the user terminal, so as to display the guidance and abetting corpus.
- the relevant personnel confirms that the instructing corpus is the real guiding and instructing corpus, it is determined that the guiding and instigating corpus is the real guiding and instructing corpus.
- the first knowledge graph may also be stored in a node of a blockchain.
- the blockchain referred to in this application is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.
- Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block.
- the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.
- the present application can be applied in the field of smart government affairs/education, and specifically in the smart supervision of smart government affairs/smart education, so as to promote the construction of smart cities.
- the aforementioned storage medium may be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only storage memory (Read-Only Memory, ROM), or a random access memory (Random Access Memory, RAM), etc.
- the present application provides an embodiment of a detection device for guiding and abetting corpus based on a knowledge graph, which is the same as the method embodiment shown in FIG. 2 .
- the apparatus can be specifically applied to various electronic devices.
- the detection device 300 for guiding and abetting corpus based on knowledge graph includes: a receiving module 301 , a building module 302 , an identification module 303 , an output module 304 and an updating module 305 .
- the receiving module 301 is used for receiving the standard corpus data set, and performing feature extraction on the standard corpus data set to obtain the standard corpus features, wherein, there is no guidance and instigation information in the standard corpus data set;
- the building module 302 is used for A first knowledge graph is constructed based on the features of the standard corpus;
- the identification module 303 is configured to receive the corpus to be detected, perform named entity recognition on the corpus to be detected, obtain the entity to be detected, and identify the entities to be detected in the first knowledge graph respectively.
- the output module 304 is configured to, when the deduction result is a deduction failure, take the entity to be detected that the deduction failed as a guiding and abetting entity, and assign the guiding and abetting entity corresponding to the deduction and abetting entity.
- the described corpus to be detected is used as the guiding and instigating corpus, and the described guiding and instigating corpus is output; the update module 305 is used to update the first knowledge map based on the successfully deduced entity to be detected when the deduction result is successful, and obtain The second knowledge graph.
- the present application proposes to detect the corpus to be detected based on the first knowledge graph, so as to determine whether the corpus to be detected belongs to the guidance and instigation corpus. Effectively realize the detection of the guidance and abetting behavior of agents in practical applications.
- the first knowledge map of the updated and expanded entities to be detected it is conducive to the continuous updating and learning of the model with the changes of the times, thereby enhancing the deterrence of the agents, further reducing the customer complaint rate, and effectively restricting the normative terms of the agents. Thereby increasing customer satisfaction.
- the above receiving module 301 is further configured to: extract triple data of each corpus in the standard corpus data set as the standard corpus feature.
- the receiving module 301 includes a word segmentation sub-module, a recognition sub-module, a determination sub-module and a screening sub-module.
- the word segmentation sub-module is used to perform word segmentation on each corpus in the standard corpus data set to obtain standard corpus words;
- the recognition sub-module is used to perform named entity recognition on the standard corpus words based on a preset entity recognition tool , to obtain a named entity set;
- the determination sub-module is used to determine the connection relationship between different named entities in the named entity set, and generate triple data based on the connection relationship;
- the screening sub-module is used to The triplet data is filtered to obtain target triplet data, and the target triplet data is used as the standard corpus feature.
- the above-mentioned building module 302 is further configured to: build the first knowledge graph based on a preset graph database and the standard corpus feature.
- the output module 304 includes a generating sub-module, a judging sub-module and a contradiction sub-module.
- the generation sub-module is used for, when the deduction result is deduction failure, the to-be-detected entity that fails to be deduced is used as the guiding and abetting entity, and the to-be-detected knowledge graph is generated based on the to-be-detected corpus corresponding to the guiding and instigating entity;
- the judgment sub-module It is used to determine whether there is a contradiction between the knowledge map to be detected and the first knowledge map;
- the contradiction sub-module is used for when there is a contradiction between the knowledge map to be detected and the first knowledge map,
- the corpus to be detected corresponding to the guiding and abetting entity is used as the guiding and instigating corpus.
- the update module 305 includes an initial qualifying submodule, a target qualifying submodule, and an update submodule.
- the initial qualified sub-module is used to identify the to-be-detected corpus corresponding to the successfully deduced entity to be detected when the deduction result is a successful deduction, as the initial qualified corpus;
- the target qualified sub-module is used when all the initially qualified corpus When all the entities to be detected are successfully deduced, the initial qualified corpus is used as the target qualified corpus;
- the update sub-module is used to update the first knowledge graph based on the target qualified corpus to obtain a second knowledge graph.
- the above-mentioned apparatus 300 further includes: a verification module, configured to verify whether the guidance and instigation corpus is a real guidance and instigation corpus, when the guidance and instigation corpus is an unauthentic guidance and instigation corpus , adding the guiding and abetting corpus to the first knowledge graph to obtain an expanded knowledge graph.
- a verification module configured to verify whether the guidance and instigation corpus is a real guidance and instigation corpus, when the guidance and instigation corpus is an unauthentic guidance and instigation corpus , adding the guiding and abetting corpus to the first knowledge graph to obtain an expanded knowledge graph.
- the above verification module is further configured to: based on a pre-trained instructing corpus detection model, detect whether the guiding and instructing corpus is a real guiding and instructing corpus.
- the verification module includes a display submodule, a request submodule, and a signal reception submodule.
- the display sub-module is used for outputting the guiding and instigating corpus to the display device of the user terminal;
- the requesting sub-module is used for outputting a signal requesting confirmation of the instigating corpus to the user terminal;
- the signal receiving sub-module is used for receiving the said
- a confirmation signal is sent by the user terminal, it is determined based on the confirmation signal whether the guiding and abetting corpus is a real guiding and instructing corpus, wherein the confirmation signal corresponds to the signal for requesting confirmation of the instructing corpus.
- the present application proposes to detect the corpus to be detected based on the first knowledge graph, so as to determine whether the corpus to be detected belongs to the guidance and instigation corpus. Effectively realize the detection of the guidance and abetting behavior of agents in practical applications.
- by deriving the first knowledge map of the updated and expanded entities to be detected it is conducive to the continuous updating and learning of the model with the changes of the times, thereby enhancing the deterrence of the agents, further reducing the customer complaint rate, and effectively restricting the normative terms of the agents. Thereby increasing customer satisfaction.
- the present application proposes to detect the corpus to be detected based on the first knowledge graph, so as to determine whether the corpus to be detected belongs to the guidance and instigation corpus. Effectively realize the detection of the guidance and abetting behavior of agents in practical applications.
- by deriving the first knowledge map of the updated and expanded entities to be detected it is conducive to the continuous updating and learning of the model with the changes of the times, thereby enhancing the deterrence of the agents, further reducing the customer complaint rate, and effectively restricting the normative terms of the agents. Thereby increasing customer satisfaction.
- FIG. 4 is a block diagram of a basic structure of a computer device according to this embodiment.
- the computer device 200 includes a memory 201 , a processor 202 , and a network interface 203 that communicate with each other through a system bus. It should be noted that only the computer device 200 with components 201-203 is shown in the figure, but it should be understood that implementation of all shown components is not required, and more or less components may be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculation and/or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, special-purpose Integrated circuit (Application Specific Integrated Circuit, ASIC), programmable gate array (Field-Programmable Gate Array, FPGA), digital processor (Digital Signal Processor, DSP), embedded equipment, etc.
- ASIC Application Specific Integrated Circuit
- FPGA Field-Programmable Gate Array
- DSP Digital Signal Processor
- the computer equipment may be a desktop computer, a notebook computer, a palmtop computer, a cloud server and other computing equipment.
- the computer device can perform human-computer interaction with the user through a keyboard, a mouse, a remote control, a touch pad or a voice control device.
- the memory 201 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static Random Access Memory (SRAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Programmable Read Only Memory (PROM), Magnetic Memory, Magnetic Disk, Optical Disk, etc.
- the computer-readable storage medium may be non-volatile or volatile.
- the memory 201 may be an internal storage unit of the computer device 200 , such as a hard disk or a memory of the computer device 200 .
- the memory 201 may also be an external storage device of the computer device 200, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, flash memory card (Flash Card), etc.
- the memory 201 may also include both an internal storage unit of the computer device 200 and an external storage device thereof.
- the memory 201 is generally used to store the operating system and various application software installed in the computer device 200 , such as computer-readable instructions for the detection method of the guidance and abetment corpus based on the knowledge graph.
- the memory 201 can also be used to temporarily store various types of data that have been output or will be output.
- the processor 202 may be a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, a microprocessor, or other data processing chips.
- the processor 202 is typically used to control the overall operation of the computer device 200 .
- the processor 202 is configured to execute computer-readable instructions stored in the memory 201 or process data, for example, computer-readable instructions for executing the knowledge graph-based method for detecting guidance and abetting corpus.
- the network interface 203 may include a wireless network interface or a wired network interface, and the network interface 203 is generally used to establish a communication connection between the computer device 200 and other electronic devices.
- the present application detects the corpus to be detected based on the first knowledge graph, so as to determine whether the corpus to be detected belongs to the guidance and instigation corpus.
- the detection of the guidance and abetting behavior of the agent in practical application is effectively realized. Effectively constrain the normative language of agents to improve customer satisfaction.
- the present application also provides another embodiment, that is, to provide a computer-readable storage medium, where the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to The at least one processor is caused to execute the steps of the above-mentioned method for detecting guidance and abetting corpus based on the knowledge graph.
- the present application detects the corpus to be detected based on the first knowledge graph, so as to determine whether the corpus to be detected belongs to the guidance and instigation corpus.
- the detection of the guidance and abetting behavior of the agent in practical application is effectively realized. Effectively constrain the normative language of agents to improve customer satisfaction.
- the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform, and of course hardware can also be used, but in many cases the former is better implementation.
- the technical solution of the present application can be embodied in the form of a software product in essence or in a part that contributes to the prior art, and the computer software product is stored in a storage medium (such as ROM/RAM, magnetic disk, CD-ROM), including several instructions to make a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) execute the methods described in the various embodiments of this application.
- a storage medium such as ROM/RAM, magnetic disk, CD-ROM
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
一种基于知识图谱的引导教唆语料的检测方法及其相关设备,包括接收标准语料数据集,对标准语料数据集进行特征提取,获得标准语料特征,标准语料数据集中不存在引导教唆信息;基于标准语料特征构建第一知识图谱;接收待检测语料,对待检测语料进行命名实体识别,获得待检测实体,在第一知识图谱中分别对每个待检测实体进行推演;当待检测实体推演失败时,将推演失败的待检测实体作为引导教唆实体,并将引导教唆实体对应的待检测语料作为引导教唆语料,输出引导教唆语料。其中,第一知识图谱可存储于区块链中。本方法能够快速识别出引导教唆语料,从而实现对引导教唆行为的检测。
Description
本申请要求于2020年12月16日提交中国专利局、申请号为202011491853.1,发明名称为“基于知识图谱的引导教唆语料的检测方法及其相关设备”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本申请涉及大数据技术领域,尤其涉及基于知识图谱的引导教唆语料的检测方法及其相关设备。
随着计算机技术的不断革新和发展,计算机已经应用到各行各业中。在坐席与客户沟通的过程中,常常容易出现引导教唆客户的情况,故而,引导教唆是语音质检中的一个常见违规场景,该违规现象出现频率高,违规性质是较为严重的,是语音质检环节中一个重要的质检点。
发明人意识到,传统质检算法多基于正则匹配规则,存在覆盖场景相对单一,泛化能力交叉的局限。同时随着坐席话术的不断优化,新兴科技的不断更新,坐席人员在引导客户方面会更加的具有创新性和时代性,导致语料数据不断的变化。若是采用完全基于规则的算法进行检测,需要耗费巨大的人力采集标注引导教唆的违规话术,以及编写冗长且复杂的规则逻辑,计算机无法随着时间的推移自学习式更新优化。
发明内容
本申请实施例的目的在于提出一种基于知识图谱的引导教唆语料的检测方法及其相关设备,快速确定出待检测语料是否属于引导教唆语料,有效实现对引导教唆行为的检测。
为了解决上述技术问题,本申请实施例提供一种基于知识图谱的引导教唆语料的检测方法,采用了如下所述的技术方案:
一种基于知识图谱的引导教唆语料的检测方法,包括下述步骤:
接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;
基于所述标准语料特征构建第一知识图谱;
接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;
当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;
当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
为了解决上述技术问题,本申请实施例还提供一种基于知识图谱的引导教唆语料的检测装置,采用了如下所述的技术方案:
一种基于知识图谱的引导教唆语料的检测装置,包括:
接收模块,用于接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;
构建模块,用于基于所述标准语料特征构建第一知识图谱;
识别模型,用于接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;
输出模块,用于当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;
更新模块,用于当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
为了解决上述技术问题,本申请实施例还提供一种计算机设备,采用了如下所述的技术方案:
一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机可读指令,所述处理器执行所述计算机可读指令时实现如下所述的基于知识图谱的引导教唆语料的检测方法的步骤:
接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;
基于所述标准语料特征构建第一知识图谱;
接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;
当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;
当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
为了解决上述技术问题,本申请实施例还提供一种计算机可读存储介质,采用了如下所述的技术方案:
一种计算机可读存储介质,所述计算机可读存储介质上存储有计算机可读指令,所述计算机可读指令被处理器执行时实现如下所述的基于知识图谱的引导教唆语料的检测方法的步骤:
接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;
基于所述标准语料特征构建第一知识图谱;
接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;
当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;
当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
与现有技术相比,本申请实施例主要有以下有益效果:
本申请提出基于第一知识图谱对待检测语料进行检测,从而确定该待检测语料是否属于引导教唆语料。有效实现对实际应用中坐席人员的引导教唆行为的检测。同时通过推演成功的待检测实体更新扩张的第一知识图谱,有利于模型随时代变迁不断更新学习,从而增强对于坐席人员的威慑性,进一步降低客户投诉率,有效的约束坐席人员的规范用语,从而提高客户满意度。
为了更清楚地说明本申请中的方案,下面将对本申请实施例描述中所需要使用的附图作一个简单介绍,显而易见地,下面描述中的附图是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1是本申请可以应用于其中的示例性系统架构图;
图2是根据本申请的基于知识图谱的引导教唆语料的检测方法的一个实施例的流程图;
图3是根据本申请的基于知识图谱的引导教唆语料的检测装置的一个实施例的结构示意图;
图4是根据本申请的计算机设备的一个实施例的结构示意图。
附图标记:200、计算机设备;201、存储器;202、处理器;203、网络接口;300、基于知识图谱的引导教唆语料的检测装置;301、接收模块;302、构建模块;303、识别模块;304、输出模块;305、更新模块。
除非另有定义,本文所使用的所有的技术和科学术语与属于本申请的技术领域的技术人员通常理解的含义相同;本文中在申请的说明书中所使用的术语只是为了描述具体的实施例的目的,不是旨在于限制本申请;本申请的说明书和权利要求书及上述附图说明中的术语“包括”和“具有”以及它们的任何变形,意图在于覆盖不排他的包含。本申请的说明书和权利要求书或上述附图中的术语“第一”、“第二”等是用于区别不同对象,而不是用于描述特定顺序。
在本文中提及“实施例”意味着,结合实施例描述的特定特征、结构或特性可以包含在本申请的至少一个实施例中。在说明书中的各个位置出现该短语并不一定均是指相同的实施例,也不是与其它实施例互斥的独立的或备选的实施例。本领域技术人员显式地和隐式地理解的是,本文所描述的实施例可以与其它实施例相结合。
为了使本技术领域的人员更好地理解本申请方案,下面将结合附图,对本申请实施例中的技术方案进行清楚、完整地描述。
如图1所示,系统架构100可以包括终端设备101、102、103,网络104和服务器105。网络104用以在终端设备101、102、103和服务器105之间提供通信链路的介质。网络104可以包括各种连接类型,例如有线、无线通信链路或者光纤电缆等等。
用户可以使用终端设备101、102、103通过网络104与服务器105交互,以接收或发送消息等。终端设备101、102、103上可以安装有各种通讯客户端应用,例如网页浏览器应用、购物类应用、搜索类应用、即时通信工具、邮箱客户端、社交平台软件等。
终端设备101、102、103可以是具有显示屏并且支持网页浏览的各种电子设备,包括但不限于智能手机、平板电脑、电子书阅读器、MP3播放器(Moving Picture Experts Group Audio Layer III,动态影像专家压缩标准音频层面3)、MP4(Moving Picture Experts Group Audio Layer IV,动态影像专家压缩标准音频层面4)播放器、膝上型便携计算机和台式计算机等等。
服务器105可以是提供各种服务的服务器,例如对终端设备101、102、103上显示的页面提供支持的后台服务器。
需要说明的是,本申请实施例所提供的基于知识图谱的引导教唆语料的检测方法一般由服务器/终端设备执行,相应地,基于知识图谱的引导教唆语料的检测装置一般设置于服务器/终端设备中。
应该理解,图1中的终端设备、网络和服务器的数目仅仅是示意性的。根据实现需要,可以具有任意数目的终端设备、网络和服务器。
继续参考图2,示出了根据本申请的基于知识图谱的引导教唆语料的检测方法的一个实施例的流程图。所述的基于知识图谱的引导教唆语料的检测方法,包括以下步骤:
S1:接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息。
在本实施例中,本申请中的标注语料数据集指不存在引导教唆信息的语料数据集,即属于合规语料。通过提取标准语料数据集中的标准语料特征,从而便于根据标准语料特征进行后续操作。
在本实施例中,基于知识图谱的引导教唆语料的检测方法运行于其上的电子设备(例如图1所示的服务器/终端设备)可以通过有线连接方式或者无线连接方式接收标注语料数据集。需要指出的是,上述无线连接方式可以包括但不限于3G/4G连接、WiFi连接、蓝牙连接、WiMAX连接、Zigbee连接、UWB(ultra wideband)连接、以及其他现在已知或将 来开发的无线连接方式。
具体的,所述对所述标准语料数据集进行提取,获得标准语料特征的步骤包括:
提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征。
在本实施例中,针对不存在引导教唆的标准语料数据集,抽取标准语料数据集中每条语料的SPO(Subject-Predicate-Object,主语-谓语-宾语)三元组数据,获得多个三元组数据,根据多个三元组数据生成三元组数据集合,作为标准语料特征。本申请将三元组数据作为标准语料特征,便于后续第一知识图谱的构建。
其中,所述提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征的步骤包括:
对所述标准语料数据集中的每条语料均进行分词操作,获得标准语料词语;
基于预设的实体识别工具对所述标准语料词语进行命名实体识别,获得命名实体集合;
确定命名实体集合中不同命名实体之间的连接关系,基于所述连接关系生成三元组数据;
基于预设的限定关系对所述三元组数据进行筛选,获得目标三元组数据,将所述目标三元组数据作为所述标准语料特征。
在本实施例中,本申请中的实体识别工具指jiagu(甲骨)。jiagu(甲骨)为深度学习自然语言处理工具,同时兼具中文分词、词性标注以及命名实体识别的功能。其中,Jiagu以BiLSTM(Bi-directional Long Short-Term Memory,双向长短期记忆)模型为基础,使用大规模语料训练而成。通过jiagu对标准语料数据集进行分词操作,获得标准语料词语。再通过jiagu对所述标准语料词语进行命名实体识别,获得命名实体集合。分词操作举例如下:原本的语料为张先是个可爱的中国人,进行分词操作后,变为['张先','是','个','可爱','的','中国人']。通过命名实体识别后,获得命名实体集合[张先,中国人]。确定不同命名实体之间的连接关系,例如,命名实体“张先”和“中国人”之间的连接关键词是“是”,则连接关系属于从属关系,则三元组数据为张先-是-中国人。基于预设的限定关系,连接所述命名实体集合中符合所述限定关系的命名实体,获得三元组数据。本申请中的限定关系可以包括父母子女关系,从属关系等常识性的关系。三元组数据举例如下:西安电子科技大学-坐标-西安;西安电子科技大学-学校类型-985工程;张某某-学历-研究生。由于标准语料数据集中不存在引导教唆的数据,生成的标准语料特征属于非引导教唆的特征。
需要说明的是,本申请根据实际需要也可以选用jieba(结巴)分词工具,适用即可。
S2:基于所述标准语料特征构建第一知识图谱。
在本实施例中,基于标准语料特征构建第一知识图谱,第一知识图谱为合规话术的知识图谱。具体步骤包括:将不同的SPO三元组之间相同的主语和/或宾语进行重合。具体的重合方式可以为主语-主语之间的重合,主语-宾语之间的重合,宾语-宾语之间的重合。
具体的,所述基于所述标准语料特征构建第一知识图谱的步骤包括
基于预设的图数据库与所述标准语料特征构建所述第一知识图谱。
在本实施例中,本申请的图数据库为Neo4j库,Neo4j库创建的图是用顶点和边构建一个有向图。利用Neo4j库与上述标准语料特征(即抽取到三元组)构建第一知识图谱,该第一知识图谱即为不涉及引导教唆数据的知识图谱。通过Neo4j库建立的第一知识图谱可以便于后续的更新扩张,本申请生成可扩张的知识图谱,有利于计算机随时代变迁不断更新学习,
S3:接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果。
在本实施例中,预测阶段,接收待质检任务,待质检任务中包括待检测语料。利用jiagu库对所述待检测语料进行分词和命名实体识别,获得待检测实体集合,并通过所述第一知识图谱历遍所述待检测实体集合中的每个待检测实体,识别所述待检测实体是否能在所述 知识图谱中进行推演。具体的推演过程为:寻找待检测实体在第一知识图谱中的路径。例如第一知识图谱中包含这样的一条路径“梅琳达·盖茨-配偶-比尔·盖茨-主席-微软-总部-西雅图”,待检测实体为梅琳达·盖茨。通过在第一知识图谱中的推演,从而获得梅林达·盖茨居住在西雅图,输出推演结果为推演成功。当待检测实体为总统时,通过在第一知识图谱中进行查找确定,无该实体。进而触发相似度算法,通过计算第一知识图谱中的各目标实体与待检测实体“物品”之间的语义相似度,确定出语义相似度超过预设的阈值的目标实体,作为替代实体,找到在第一知识图谱中替代实体的路径,从而确定推演结果为推演成功。若第一知识图谱中不存在与待检测实体“物品”之间的语义相似度超过预设的阈值的目标实体,则输出的推演结果为推演失败。
需要说明的是:本申请包括但不限于上述推演过程,在实际的应用过程中,可以根据实际需要选用任意一种推演方式,适用即可。
S4:当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料。
在本实施例中,通过推演失败的待检测实体确定出对应的待检测语料,从而确定出引导教唆语料。实现对引导教唆语料的快速识别。从而有效约束坐席人员的规范用语,降低客户投诉率,提高客户满意度。本申请同时可以将引导教唆语料对应的场景作为引导教唆场景。
具体的,当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的待检测语料作为引导教唆语料,输出所述引导教唆语料的步骤包括:
当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并基于所述引导教唆实体对应的待检测语料生成待检测知识图谱;
确定所述待检测知识图谱和所述第一知识图谱之间是否存在矛盾关系;
当所述待检测知识图谱和所述第一知识图谱之间存在矛盾关系时,将所述引导教唆实体对应的待检测语料作为引导教唆语料。
在本实施例中,提取待检测语料中的三元组数据,作为待检测三元组数据。基于待检测三元组数据构建待检测知识图谱。无法在第一知识图谱上推演的待检测实体为引导教唆实体,然后依据实体间空间位置关系判断待检测语料是否为引导教唆语料,具体为对比待检测知识图谱和第一知识图谱是否存在矛盾关系。若存在矛盾关系,则确定该待检测语料为引导教唆语料。该待检测语料所对应的场景为引导教唆场景。若不存在矛盾关系,则将该待检测语料作为待确认语料,保存到预设的数据库中。其中,推演可以正向推演,即从主语向宾语的方向推演,也可以反向推演,即从宾语向主语的方向推演。对于实体的主语和宾语的判断,计算机在通过jiagu库进行分词的过程中,已经确定每个词语的词性,并对每个词语进行词性标注,即标注每个词语是属于主语、宾语、谓语或者形容词性定语等。
本申请中的矛盾关系是指不同的知识图谱的相互之间的逻辑上的表达冲突关系,例如,在待检测知识图谱中存在“张先的学历-是-小学”的三元组数据,而在第一知识图谱中却存在“张先的学历-是-研究生”的三元组数据,则此时,不同知识图谱中的三元组数据是矛盾的,进而确定待检测知识图谱与第一知识图谱之间存在矛盾关系。
S5:当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
在本实施例中,通过推演成功的待检测实体更新扩充第一知识图谱,实现知识图谱的不断更新,进而实现计算机对引导教唆语料的自学习式的更新优化。
具体的,当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱的步骤包括:
当所述推演结果为推演成功时,识别推演成功的待检测实体对应的待检测语料,作为初始合格语料;
当所述初始合格语料中所有的待检测实体均推演成功时,将所述初始合格语料作为目标合格语料;
基于所述目标合格语料更新所述第一知识图谱,获得第二知识图谱。
在本实施例中,基于目标合格语料更新所述第一知识图谱,获得第二知识图谱的具体步骤包括:将目标合格语料转换为三元组数据,将该三元组数据增加到第一知识图谱中,获得第二知识图谱。通过任意一个推演成功的实体,快速确定出初始合格语料,通过判断初始合格语料中的实体是否全部推演成功,当初始合格语料中的全部实体均推演成功时,则可以直接将该初始合格语料作为目标合格语料,实现快速确定出目标合格语料。
在本实施例的一些可选的实现方式中,在步骤S4,即将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料之后,上述电子设备还可以执行以下步骤:
验证所述引导教唆语料是否为真实引导教唆语料,在所述引导教唆语料为非真实引导教唆语料时,将所述引导教唆语料增加到所述第一知识图谱中,获得扩张知识图谱。
在本实施例中,将验证之后,确定出的引导教唆语料不是真实的引导教唆语料时,则认为该引导教唆语料实际上是合规语料,将该语料添加至第一知识图谱中,实现第一知识图谱的扩张。即复核上述判断语料属于引导教唆语料的质检结果,对于不违规的场景,将该知识增加到第一知识图谱中。
具体的,所述验证所述引导教唆语料是否为真实引导教唆语料的步骤包括:
基于预先训练的教唆语料检测模型检测所述引导教唆语料是否为真实引导教唆语料。
在本实施例中,在知识图谱已经确定对应的语料为引导教唆语料的前提下,通过预先训练的教唆语料检测模型对所述引导教唆语料进行二次检测验证。若此时教唆语料检测模型输出的结果是所述引导教唆语料为真实引导教唆语料,则更加可以确定该语料属于引导教唆型的语料。进而更加可以确定该引导教唆型的语料所对应的场景,属于引导教唆型的场景。本申请的教唆语料检测模型为NLP(Natural Language Processing,自然语言处理)模型。
此外,作为本申请的另一实施例,所述验证所述引导教唆语料是否为真实引导教唆语料的步骤包括:
将所述引导教唆语料输出至用户终端的显示设备;
向所述用户终端输出请求确认教唆语料的信号;
当接收到所述用户终端发送的确认信号时,基于所述确认信号确定所述引导教唆语料是否为真实引导教唆语料,其中,所述确认信号与所述请求确认教唆语料的信号相对应。
在本实施例中,将所述引导教唆语料输出至用户终端的显示设备,以进行所述引导教唆语料的展示。当所述相关人员确认所述教唆语料为真实引导教唆语料时,确定所述引导教唆语料为真实引导教唆语料。
需要强调的是,为进一步保证上述第一知识图谱的私密和安全性,第一知识图谱还可以存储于一区块链的节点中。
本申请所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain),本质上是一个去中心化的数据库,是一串使用密码学方法相关联产生的数据块,每一个数据块中包含了一批次网络交易的信息,用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
本申请可应用于智慧政务/教育领域领域中,具体可以应用于智慧政务/智慧教育的智慧监管中,从而推动智慧城市的建设。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机可读指令来指令相关的硬件来完成,该计算机可读指令可存储于一计算机可读取存储介质中,该计算机可读指令在执行时,可包括如上述各方法的实施例的流程。其中,前 述的存储介质可为磁碟、光盘、只读存储记忆体(Read-Only Memory,ROM)等非易失性存储介质,或随机存储记忆体(Random Access Memory,RAM)等。
应该理解的是,虽然附图的流程图中的各个步骤按照箭头的指示依次显示,但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明,这些步骤的执行并没有严格的顺序限制,其可以以其他的顺序执行。而且,附图的流程图中的至少一部分步骤可以包括多个子步骤或者多个阶段,这些子步骤或者阶段并不必然是在同一时刻执行完成,而是可以在不同的时刻执行,其执行顺序也不必然是依次进行,而是可以与其他步骤或者其他步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。
进一步参考图3,作为对上述图2所示方法的实现,本申请提供了一种基于知识图谱的引导教唆语料的检测装置的一个实施例,该装置实施例与图2所示的方法实施例相对应,该装置具体可以应用于各种电子设备中。
如图3所示,本实施例所述的基于知识图谱的引导教唆语料的检测装置300包括:接收模块301、构建模块302、识别模块303、输出模块304以及更新模块305。其中:接收模块301,用于接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;构建模块302,用于基于所述标准语料特征构建第一知识图谱;识别模块303,用于接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;输出模块304,用于当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;更新模块305,用于当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
在本实施例中,本申请提出基于第一知识图谱对待检测语料进行检测,从而确定该待检测语料是否属于引导教唆语料。有效实现对实际应用中坐席人员的引导教唆行为的检测。同时通过推演成功的待检测实体更新扩张的第一知识图谱,有利于模型随时代变迁不断更新学习,从而增强对于坐席人员的威慑性,进一步降低客户投诉率,有效的约束坐席人员的规范用语,从而提高客户满意度。
在本实施例的一些可选的实现方式中,上述接收模块301进一步用于:提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征。
接收模块301包括分词子模块、识别子模块、确定子模块和筛选子模块。其中,分词子模块用于对所述标准语料数据集中的每条语料均进行分词操作,获得标准语料词语;识别子模块用于基于预设的实体识别工具对所述标准语料词语进行命名实体识别,获得命名实体集合;确定子模块用于确定命名实体集合中不同命名实体之间的连接关系,基于所述连接关系生成三元组数据;筛选子模块用于基于预设的限定关系对所述三元组数据进行筛选,获得目标三元组数据,将所述目标三元组数据作为所述标准语料特征。
在本实施例的一些可选的实现方式中,上述构建模块302进一步用于:基于预设的图数据库与所述标准语料特征构建所述第一知识图谱。
输出模块304包括生成子模块、判断子模块和矛盾子模块。其中,生成子模块用于当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并基于所述引导教唆实体对应的待检测语料生成待检测知识图谱;判断子模块用于确定所述待检测知识图谱和所述第一知识图谱之间是否存在矛盾关系;矛盾子模块用于当所述待检测知识图谱和所述第一知识图谱之间存在矛盾关系时,将所述引导教唆实体对应的待检测语料作为引导教唆语料。
更新模块305包括初始合格子模块、目标合格子模块和更新子模块。其中,初始合格子模块用于当所述推演结果为推演成功时,识别推演成功的待检测实体对应的待检测语料,作为初始合格语料;目标合格子模块用于当所述初始合格语料中所有的待检测实体均推演 成功时,将所述初始合格语料作为目标合格语料;更新子模块用于基于所述目标合格语料更新所述第一知识图谱,获得第二知识图谱。
在本实施例的一些可选的实现方式中,上述装置300还包括:验证模块,用于验证所述引导教唆语料是否为真实引导教唆语料,在所述引导教唆语料为非真实引导教唆语料时,将所述引导教唆语料增加到所述第一知识图谱中,获得扩张知识图谱。
在本实施例的一些可选的实现方式中,上述验证模块进一步用于:基于预先训练的教唆语料检测模型检测所述引导教唆语料是否为真实引导教唆语料。
在本实施例的一些可选的实现方式中,验证模块包括显示子模块、请求子模块和信号接收子模块。其中,显示子模块用于将所述引导教唆语料输出至用户终端的显示设备;请求子模块用于向所述用户终端输出请求确认教唆语料的信号;信号接收子模块用于当接收到所述用户终端发送的确认信号时,基于所述确认信号确定所述引导教唆语料是否为真实引导教唆语料,其中,所述确认信号与所述请求确认教唆语料的信号相对应。
本申请提出基于第一知识图谱对待检测语料进行检测,从而确定该待检测语料是否属于引导教唆语料。有效实现对实际应用中坐席人员的引导教唆行为的检测。同时通过推演成功的待检测实体更新扩张的第一知识图谱,有利于模型随时代变迁不断更新学习,从而增强对于坐席人员的威慑性,进一步降低客户投诉率,有效的约束坐席人员的规范用语,从而提高客户满意度。
本申请提出基于第一知识图谱对待检测语料进行检测,从而确定该待检测语料是否属于引导教唆语料。有效实现对实际应用中坐席人员的引导教唆行为的检测。同时通过推演成功的待检测实体更新扩张的第一知识图谱,有利于模型随时代变迁不断更新学习,从而增强对于坐席人员的威慑性,进一步降低客户投诉率,有效的约束坐席人员的规范用语,从而提高客户满意度。
为解决上述技术问题,本申请实施例还提供计算机设备。具体请参阅图4,图4为本实施例计算机设备基本结构框图。
所述计算机设备200包括通过系统总线相互通信连接存储器201、处理器202、网络接口203。需要指出的是,图中仅示出了具有组件201-203的计算机设备200,但是应理解的是,并不要求实施所有示出的组件,可以替代的实施更多或者更少的组件。其中,本技术领域技术人员可以理解,这里的计算机设备是一种能够按照事先设定或存储的指令,自动进行数值计算和/或信息处理的设备,其硬件包括但不限于微处理器、专用集成电路(Application Specific Integrated Circuit,ASIC)、可编程门阵列(Field-Programmable Gate Array,FPGA)、数字处理器(Digital Signal Processor,DSP)、嵌入式设备等。
所述计算机设备可以是桌上型计算机、笔记本、掌上电脑及云端服务器等计算设备。所述计算机设备可以与用户通过键盘、鼠标、遥控器、触摸板或声控设备等方式进行人机交互。
所述存储器201至少包括一种类型的可读存储介质,所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如,SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器(PROM)、磁性存储器、磁盘、光盘等。所述计算机可读存储介质可以是非易失性,也可以是易失性。在一些实施例中,所述存储器201可以是所述计算机设备200的内部存储单元,例如该计算机设备200的硬盘或内存。在另一些实施例中,所述存储器201也可以是所述计算机设备200的外部存储设备,例如该计算机设备200上配备的插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)等。当然,所述存储器201还可以既包括所述计算机设备200的内部存储单元也包括其外部存储设备。本实施例中,所述存储器201通常用于存储安装于所述计算机设备200的操作系统和各类应用软件,例如基于知识图谱的引导教唆语料的检测方法的计算机可读指令等。此外,所述存储器201还可以用于暂时地存储已经输出或者将 要输出的各类数据。
所述处理器202在一些实施例中可以是中央处理器(Central Processing Unit,CPU)、控制器、微控制器、微处理器、或其他数据处理芯片。该处理器202通常用于控制所述计算机设备200的总体操作。本实施例中,所述处理器202用于运行所述存储器201中存储的计算机可读指令或者处理数据,例如运行所述基于知识图谱的引导教唆语料的检测方法的计算机可读指令。
所述网络接口203可包括无线网络接口或有线网络接口,该网络接口203通常用于在所述计算机设备200与其他电子设备之间建立通信连接。
在本实施例中,本申请基于第一知识图谱对待检测语料进行检测,从而确定该待检测语料是否属于引导教唆语料。从而有效实现对实际应用中坐席人员的引导教唆行为的检测。有效的约束坐席人员的规范用语,提高客户满意度。
本申请还提供了另一种实施方式,即提供一种计算机可读存储介质,所述计算机可读存储介质存储有计算机可读指令,所述计算机可读指令可被至少一个处理器执行,以使所述至少一个处理器执行如上述的基于知识图谱的引导教唆语料的检测方法的步骤。
在本实施例中,本申请基于第一知识图谱对待检测语料进行检测,从而确定该待检测语料是否属于引导教唆语料。从而有效实现对实际应用中坐席人员的引导教唆行为的检测。有效的约束坐席人员的规范用语,提高客户满意度。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中,包括若干指令用以使得一台终端设备(可以是手机,计算机,服务器,空调器,或者网络设备等)执行本申请各个实施例所述的方法。
显然,以上所描述的实施例仅仅是本申请一部分实施例,而不是全部的实施例,附图中给出了本申请的较佳实施例,但并不限制本申请的专利范围。本申请可以以许多不同的形式来实现,相反地,提供这些实施例的目的是使对本申请的公开内容的理解更加透彻全面。尽管参照前述实施例对本申请进行了详细的说明,对于本领域的技术人员来而言,其依然可以对前述各具体实施方式所记载的技术方案进行修改,或者对其中部分技术特征进行等效替换。凡是利用本申请说明书及附图内容所做的等效结构,直接或间接运用在其他相关的技术领域,均同理在本申请专利保护范围之内。
Claims (20)
- 一种基于知识图谱的引导教唆语料的检测方法,包括下述步骤:接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;基于所述标准语料特征构建第一知识图谱;接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
- 根据权利要求1所述的基于知识图谱的引导教唆语料的检测方法,其中,所述对所述标准语料数据集进行提取,获得标准语料特征的步骤包括:提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征。
- 根据权利要求2所述的基于知识图谱的引导教唆语料的检测方法,其中,所述提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征的步骤包括:对所述标准语料数据集中的每条语料均进行分词操作,获得标准语料词语;基于预设的实体识别工具对所述标准语料词语进行命名实体识别,获得命名实体集合;确定命名实体集合中不同命名实体之间的连接关系,基于所述连接关系生成三元组数据;基于预设的限定关系对所述三元组数据进行筛选,获得目标三元组数据,将所述目标三元组数据作为所述标准语料特征。
- 根据权利要求1所述的基于知识图谱的引导教唆语料的检测方法,其中,当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的待检测语料作为引导教唆语料,输出所述引导教唆语料的步骤包括:当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并基于所述引导教唆实体对应的待检测语料生成待检测知识图谱;确定所述待检测知识图谱和所述第一知识图谱之间是否存在矛盾关系;当所述待检测知识图谱和所述第一知识图谱之间存在矛盾关系时,将所述引导教唆实体对应的待检测语料作为引导教唆语料。
- 根据权利要求1所述的基于知识图谱的引导教唆语料的检测方法,其中,当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱的步骤包括:当所述推演结果为推演成功时,识别推演成功的待检测实体对应的待检测语料,作为初始合格语料;当所述初始合格语料中所有的待检测实体均推演成功时,将所述初始合格语料作为目标合格语料;基于所述目标合格语料更新所述第一知识图谱,获得第二知识图谱。
- 根据权利要求1所述的基于知识图谱的引导教唆语料的检测方法,其中,在所述将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料的步骤之后,还包括:验证所述引导教唆语料是否为真实引导教唆语料,在所述引导教唆语料为非真实引导教唆语料时,将所述引导教唆语料增加到所述第一知识图谱中,获得扩张知识图谱。
- 根据权利要求6所述的基于知识图谱的引导教唆语料的检测方法,其中,所述验证所述引导教唆语料是否为真实引导教唆语料的步骤包括:将所述引导教唆语料输出至用户终端的显示设备;向所述用户终端输出请求确认教唆语料的信号;当接收到所述用户终端发送的确认信号时,基于所述确认信号确定所述引导教唆语料是否为真实引导教唆语料,其中,所述确认信号与所述请求确认教唆语料的信号相对应。
- 一种基于知识图谱的引导教唆语料的检测装置,包括:接收模块,用于接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;构建模块,用于基于所述标准语料特征构建第一知识图谱;识别模块,用于接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;输出模块,用于当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;更新模块,用于当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
- 一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机可读指令,所述处理器执行所述计算机可读指令时实现如下所述的基于知识图谱的引导教唆语料的检测方法的步骤:接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;基于所述标准语料特征构建第一知识图谱;接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
- 根据权利要求9所述的计算机设备,其中,所述对所述标准语料数据集进行提取,获得标准语料特征的步骤包括:提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征。
- 根据权利要求10所述的计算机设备,其中,所述提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征的步骤包括:对所述标准语料数据集中的每条语料均进行分词操作,获得标准语料词语;基于预设的实体识别工具对所述标准语料词语进行命名实体识别,获得命名实体集合;确定命名实体集合中不同命名实体之间的连接关系,基于所述连接关系生成三元组数据;基于预设的限定关系对所述三元组数据进行筛选,获得目标三元组数据,将所述目标三元组数据作为所述标准语料特征。
- 根据权利要求9所述的计算机设备,其中,当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的待检测语料作为引导教唆语料,输出所述引导教唆语料的步骤包括:当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并基于所述引导教唆实体对应的待检测语料生成待检测知识图谱;确定所述待检测知识图谱和所述第一知识图谱之间是否存在矛盾关系;当所述待检测知识图谱和所述第一知识图谱之间存在矛盾关系时,将所述引导教唆实体对应的待检测语料作为引导教唆语料。
- 根据权利要求9所述的计算机设备,其中,当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱的步骤包括:当所述推演结果为推演成功时,识别推演成功的待检测实体对应的待检测语料,作为初始合格语料;当所述初始合格语料中所有的待检测实体均推演成功时,将所述初始合格语料作为目标合格语料;基于所述目标合格语料更新所述第一知识图谱,获得第二知识图谱。
- 根据权利要求9所述的计算机设备,其中,在所述将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料的步骤之后,还包括:验证所述引导教唆语料是否为真实引导教唆语料,在所述引导教唆语料为非真实引导教唆语料时,将所述引导教唆语料增加到所述第一知识图谱中,获得扩张知识图谱。
- 根据权利要求14所述的计算机设备,其中,所述验证所述引导教唆语料是否为真实引导教唆语料的步骤包括:将所述引导教唆语料输出至用户终端的显示设备;向所述用户终端输出请求确认教唆语料的信号;当接收到所述用户终端发送的确认信号时,基于所述确认信号确定所述引导教唆语料是否为真实引导教唆语料,其中,所述确认信号与所述请求确认教唆语料的信号相对应。
- 一种计算机可读存储介质,所述计算机可读存储介质上存储有计算机可读指令,所述计算机可读指令被处理器执行时实现如下所述的基于知识图谱的引导教唆语料的检测方法的步骤:接收标准语料数据集,对所述标准语料数据集进行特征提取,获得标准语料特征,其中,所述标准语料数据集中不存在引导教唆信息;基于所述标准语料特征构建第一知识图谱;接收待检测语料,对所述待检测语料进行命名实体识别,获得待检测实体,并在所述第一知识图谱中分别对每个所述待检测实体进行推演,获得推演结果;当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的所述待检测语料作为引导教唆语料,输出所述引导教唆语料;当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱。
- 根据权利要求16所述的计算机可读存储介质,其中,所述对所述标准语料数据集进行提取,获得标准语料特征的步骤包括:提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征。
- 根据权利要求17所述的计算机可读存储介质,其中,所述提取所述标准语料数据集中每条语料的三元组数据,作为所述标准语料特征的步骤包括:对所述标准语料数据集中的每条语料均进行分词操作,获得标准语料词语;基于预设的实体识别工具对所述标准语料词语进行命名实体识别,获得命名实体集合;确定命名实体集合中不同命名实体之间的连接关系,基于所述连接关系生成三元组数据;基于预设的限定关系对所述三元组数据进行筛选,获得目标三元组数据,将所述目标三元组数据作为所述标准语料特征。
- 根据权利要求16所述的计算机可读存储介质,其中,当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并将所述引导教唆实体对应的待检测语料作为引导教唆语料,输出所述引导教唆语料的步骤包括:当所述推演结果为推演失败时,将推演失败的待检测实体作为引导教唆实体,并基于所述引导教唆实体对应的待检测语料生成待检测知识图谱;确定所述待检测知识图谱和所述第一知识图谱之间是否存在矛盾关系;当所述待检测知识图谱和所述第一知识图谱之间存在矛盾关系时,将所述引导教唆实体对应的待检测语料作为引导教唆语料。
- 根据权利要求16所述的计算机可读存储介质,其中,当所述推演结果为推演成功时,基于推演成功的待检测实体更新所述第一知识图谱,获得第二知识图谱的步骤包括:当所述推演结果为推演成功时,识别推演成功的待检测实体对应的待检测语料,作为初始合格语料;当所述初始合格语料中所有的待检测实体均推演成功时,将所述初始合格语料作为目标合格语料;基于所述目标合格语料更新所述第一知识图谱,获得第二知识图谱。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011491853.1 | 2020-12-16 | ||
CN202011491853.1A CN112528040B (zh) | 2020-12-16 | 2020-12-16 | 基于知识图谱的引导教唆语料的检测方法及其相关设备 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022126962A1 true WO2022126962A1 (zh) | 2022-06-23 |
Family
ID=75000902
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2021/090164 WO2022126962A1 (zh) | 2020-12-16 | 2021-04-27 | 基于知识图谱的引导教唆语料的检测方法及其相关设备 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN112528040B (zh) |
WO (1) | WO2022126962A1 (zh) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117573809A (zh) * | 2024-01-12 | 2024-02-20 | 中电科大数据研究院有限公司 | 一种基于事件图谱的舆情推演方法以及相关装置 |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112528040B (zh) * | 2020-12-16 | 2024-03-19 | 平安科技(深圳)有限公司 | 基于知识图谱的引导教唆语料的检测方法及其相关设备 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180349781A1 (en) * | 2017-06-02 | 2018-12-06 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method and device for judging news quality and storage medium |
CN110688489A (zh) * | 2019-09-09 | 2020-01-14 | 中国电子科技集团公司电子科学研究院 | 基于交互注意力的知识图谱推演方法、装置和存储介质 |
CN111061843A (zh) * | 2019-12-26 | 2020-04-24 | 武汉大学 | 一种知识图谱引导的假新闻检测方法 |
CN111460167A (zh) * | 2020-03-19 | 2020-07-28 | 平安国际智慧城市科技股份有限公司 | 基于知识图谱定位排污对象的方法及相关设备 |
CN112528040A (zh) * | 2020-12-16 | 2021-03-19 | 平安科技(深圳)有限公司 | 基于知识图谱的引导教唆语料的检测方法及其相关设备 |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10681061B2 (en) * | 2017-06-14 | 2020-06-09 | International Business Machines Corporation | Feedback-based prioritized cognitive analysis |
US10938817B2 (en) * | 2018-04-05 | 2021-03-02 | Accenture Global Solutions Limited | Data security and protection system using distributed ledgers to store validated data in a knowledge graph |
CN110290116B (zh) * | 2019-06-04 | 2021-06-22 | 中山大学 | 一种基于知识图谱的恶意域名检测方法 |
CN110941664B (zh) * | 2019-12-11 | 2024-01-09 | 北京百度网讯科技有限公司 | 知识图谱的构建方法、检测方法、装置、设备及存储介质 |
-
2020
- 2020-12-16 CN CN202011491853.1A patent/CN112528040B/zh active Active
-
2021
- 2021-04-27 WO PCT/CN2021/090164 patent/WO2022126962A1/zh active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180349781A1 (en) * | 2017-06-02 | 2018-12-06 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method and device for judging news quality and storage medium |
CN110688489A (zh) * | 2019-09-09 | 2020-01-14 | 中国电子科技集团公司电子科学研究院 | 基于交互注意力的知识图谱推演方法、装置和存储介质 |
CN111061843A (zh) * | 2019-12-26 | 2020-04-24 | 武汉大学 | 一种知识图谱引导的假新闻检测方法 |
CN111460167A (zh) * | 2020-03-19 | 2020-07-28 | 平安国际智慧城市科技股份有限公司 | 基于知识图谱定位排污对象的方法及相关设备 |
CN112528040A (zh) * | 2020-12-16 | 2021-03-19 | 平安科技(深圳)有限公司 | 基于知识图谱的引导教唆语料的检测方法及其相关设备 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117573809A (zh) * | 2024-01-12 | 2024-02-20 | 中电科大数据研究院有限公司 | 一种基于事件图谱的舆情推演方法以及相关装置 |
CN117573809B (zh) * | 2024-01-12 | 2024-05-10 | 中电科大数据研究院有限公司 | 一种基于事件图谱的舆情推演方法以及相关装置 |
Also Published As
Publication number | Publication date |
---|---|
CN112528040B (zh) | 2024-03-19 |
CN112528040A (zh) | 2021-03-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2022022045A1 (zh) | 基于知识图谱的文本比对方法、装置、设备及存储介质 | |
WO2022174491A1 (zh) | 基于人工智能的病历质控方法、装置、计算机设备及存储介质 | |
CN109428886B (zh) | 用于经由区块链进行评论验证和可信度评分的方法和系统 | |
US9626622B2 (en) | Training a question/answer system using answer keys based on forum content | |
CN108090351B (zh) | 用于处理请求消息的方法和装置 | |
US20180365773A1 (en) | Anti-money laundering platform for mining and analyzing data to identify money launderers | |
US20160085740A1 (en) | Generating training data for disambiguation | |
US10489127B2 (en) | Mapping of software code via user interface summarization | |
WO2023134057A1 (zh) | 事务信息查询方法、装置、计算机设备及存储介质 | |
WO2022105119A1 (zh) | 意图识别模型的训练语料生成方法及其相关设备 | |
WO2022126962A1 (zh) | 基于知识图谱的引导教唆语料的检测方法及其相关设备 | |
US8990138B2 (en) | Automated verification of hypotheses using ontologies | |
US11954173B2 (en) | Data processing method, electronic device and computer program product | |
CN112084342A (zh) | 试题生成方法、装置、计算机设备及存储介质 | |
CN112417887A (zh) | 敏感词句识别模型处理方法、及其相关设备 | |
CN112085087A (zh) | 业务规则生成的方法、装置、计算机设备及存储介质 | |
CN110618999A (zh) | 数据的查询方法及装置、计算机存储介质、电子设备 | |
US20150379112A1 (en) | Creating an on-line job function ontology | |
CN112363814A (zh) | 任务调度方法、装置、计算机设备及存储介质 | |
Soni et al. | Follow the leader: Documents on the leading edge of semantic change get more citations | |
CN115733763A (zh) | 一种关联网络的标签传播方法、装置及计算机可读存储介质 | |
WO2022073341A1 (zh) | 基于语音语义的疾病实体匹配方法、装置及计算机设备 | |
CN105354506B (zh) | 隐藏文件的方法和装置 | |
KR102135075B1 (ko) | 뉴스 작성 지침 및 방송 보도 지침 기반의 인스턴트 메시지의 구문 분석을 통한 가짜 뉴스 알림 서비스 제공 방법 및 장치 | |
CN108768742B (zh) | 网络构建方法及装置、电子设备、存储介质 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21904901 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 21904901 Country of ref document: EP Kind code of ref document: A1 |