Text information processing method, device, and electronic device

Abstract

The present application provides a text information processing method, comprising: obtaining target text information, and obtaining designated named entity information in the target text information; determining target attribute slot information corresponding to the designated named entity information according to the designated named entity information; extracting attribute information of a first target attribute from the target text information according to the designated named entity information and the target attribute slot information, and determining attribute information of a second target attribute according to the target text information, the designated named entity information and the target attribute slot information. The text information processing method provided by the present application can improve the flexibility and accuracy of entity-attribute-attribute information extraction of text information.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and more particularly to a text information processing method. The present application also relates to a text information processing device, an electronic device, and a storage medium. Background Art

[0002] With the rapid development of science and technology, relevant personnel often face the problem of how to deal with and utilize massive text information in daily life. For example, with the rapid development of Internet technology, relevant personnel often need to face the problem of how to quickly and accurately analyze truly useful information from massive network text information. For example, with the rapid development of industrial technology, more and more large and heavy equipment are used in industrial production and processing operations. Once these equipment fail, it will not only cause direct losses due to equipment damage, but also cause indirect losses in industrial production and processing operations. In order to improve the reliability of these equipment, industrial enterprises often monitor, inspect, and repair these equipment by deploying a large number of sensors or other monitoring devices. In the process of monitoring, inspecting, and repairing these equipment, a large amount of equipment quality text information related to equipment quality information is often generated, and most of the quality text information is manually written and mainly unstructured text information, including but not limited to: inspection logs, defect records, failure cases, disposal plans, technical specifications, standard guidelines, instruction manuals, instructions, scientific research papers, etc. Relevant personnel often need to face the problem of how to analyze, summarize and utilize massive quality text information.

[0003] In order to solve the problem that massive text data information cannot be fully processed and utilized, it is necessary to extract, train and model the "entity-attribute-attribute information" of massive text information based on natural language processing and knowledge graph construction technology, build a professional knowledge base for equipment, and precipitate the methods and experience in the past documents and experts' brains into the equipment knowledge graph. Therefore, how to extract "entity-attribute-attribute information" from text information has become the key to solving the problem that massive text data information cannot be fully processed and utilized. There are generally two methods for extracting "entity-attribute-attribute information" from text information. The first method is to use word distance or dependency syntax, and write extraction rules based on the summary of rules to achieve the extraction of "entity-attribute-attribute information" of text information. The disadvantage of this method is that it is inflexible and cannot be applied to situations with diverse text forms and syntax, and the extraction accuracy is low; the second method is to extract "entity-attribute-attribute information" from text information based on sequence-to-sequence models. The disadvantage of this method is that it can only extract attribute-attribute information that can be directly determined in text information. Summary of the invention

[0004] The present application provides a text information processing method, device, electronic device and storage medium to improve the flexibility and accuracy of extracting "entity-attribute-attribute information" from text information.

[0005] The present application provides a text information processing method, comprising:

[0006] Obtaining target text information, and obtaining specified named entity information in the target text information;

[0007] Determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0008] According to the specified named entity information and the target attribute slot information, attribute information of a first target attribute is extracted from the target text information, and attribute information of a second target attribute is determined according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information.

[0009] Optionally, extracting the attribute information of the first target attribute from the target text information according to the specified named entity information and the target attribute slot information includes: inputting the target text information, the specified named entity information and the target attribute slot information into a first attribute information extraction module to obtain the attribute information of the first target attribute, and the first attribute information extraction module is used to determine the attribute information of the first target attribute according to the target text information, the specified named entity information and the target attribute slot information.

[0010] Optionally, the step of inputting the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information to obtain the attribute information of the first target attribute includes:

[0011] Input the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information, and obtain keyword information corresponding to the target text information, entity feature information corresponding to the designated named entity information, feature information corresponding to the target attribute, and position feature information of the designated named entity information in the target text information;

[0012] The attribute information of the first target attribute is determined according to the keyword information corresponding to the target text information, the entity feature information corresponding to the specified named entity information, the feature information corresponding to the target attribute and the position feature information of the specified named entity information in the target text information.

[0013] Optionally, also include:

[0014] Determining attribute information of a first target sample attribute;

[0015] For the attribute information of the first target sample attribute, obtain sample text information corresponding to the attribute information of the first target sample attribute, specified named entity information in the sample text information, and attribute slot information corresponding to the specified named entity information in the sample text information;

[0016] The first attribute information extraction module is determined according to the attribute information of the first target sample attribute, the sample text information corresponding to the attribute information of the first target sample attribute, the specified named entity information in the sample text information, and the attribute slot information corresponding to the specified named entity information in the sample text information.

[0017] Optionally, the step of inputting the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information to obtain the attribute information of the first target attribute comprises: inputting the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information to obtain the first target attribute, and obtaining the attribute information of the first target attribute;

[0018] The determining the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information includes:

[0019] Determine the second target attribute according to the target attribute slot information and the first target attribute, and determine the designated named entity corresponding to the second target attribute;

[0020] The second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into a second attribute information extraction module to obtain attribute information of the second target attribute. The second attribute information extraction module is used to obtain the attribute information of the second target attribute based on the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information.

[0021] Optionally, the step of inputting the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information into a second attribute information extraction module to obtain the attribute information of the second target attribute includes:

[0022] Inputting the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information into a second attribute information extraction module to obtain feature information corresponding to the second target attribute;

[0023] determining, in the target text information, candidate attribute information matching the feature information corresponding to the second target attribute according to the feature information corresponding to the second target attribute;

[0024] Attribute information of the second target attribute is obtained according to the candidate attribute information, the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information.

[0025] Optionally, also include:

[0026] Determining attribute information of a second target sample attribute;

[0027] For the attribute information of the second target sample attribute, determine the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute;

[0028] The second attribute information extraction module is determined according to the attribute information of the second target sample attribute, the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute.

[0029] Optionally, the obtaining of target text information and obtaining designated named entity information in the target text information includes: performing named entity recognition on the target text information to obtain the designated named entity information.

[0030] Optionally, determining, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information includes:

[0031] Determining, according to the specified named entity information, type information corresponding to the specified named entity information;

[0032] Obtaining a target entity named link corresponding to the target text information, wherein the target entity named link is a link pre-set for the target text information and used to obtain target attribute slot information corresponding to the named entity information according to type information corresponding to the named entity information;

[0033] According to the type information of the designated named entity and the target entity naming link, target attribute slot information corresponding to the designated named entity information is determined.

[0034] Optionally, also include:

[0035] Obtaining attribute information of the target attribute according to the attribute information of the first target attribute and the attribute information of the second target attribute;

[0036] According to the attribute information of the target attribute and the designated named entity information, a set of the designated named entity information, the target attribute, and the attribute information of the target attribute corresponding to the target text information is determined.

[0037] In another aspect, the present application provides a text information processing device, comprising:

[0038] A named entity information obtaining unit, used to obtain target text information and obtain designated named entity information in the target text information;

[0039] An attribute slot information obtaining unit, configured to determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information indicating a target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0040] A target attribute information obtaining unit is used to extract attribute information of a first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine attribute information of a second target attribute according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information.

[0041] On the other hand, the present application provides a text information processing system, including: a text collection device, a text processing device and a text integration device, wherein the text processing device includes: a named entity information acquisition module, an attribute slot information acquisition module and a target attribute information acquisition module;

[0042] The text collection device is used to collect target text information;

[0043] The named entity information obtaining module is used to obtain target text information and obtain specified named entity information in the target text information;

[0044] The attribute slot information obtaining module is used to determine the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, wherein the target attribute slot information is information for indicating the target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0045] The target attribute information obtaining module is used to extract attribute information of a first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine attribute information of a second target attribute according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information;

[0046] The text integration device is used to obtain the attribute information of the target attribute based on the attribute information of the first target attribute and the attribute information of the second target attribute, and determine the specified named entity information corresponding to the target text information, the target attribute, and the set of the attribute information of the target attribute based on the attribute information of the target attribute and the specified named entity information.

[0047] On the other hand, the present application provides a knowledge graph construction system, including: a text processing device and a knowledge graph construction device, wherein the text processing device includes: a text collection module, a named entity information acquisition module, an attribute slot information acquisition module, a target attribute information acquisition module, and a text integration module;

[0048] The text collection module is used to collect target text information;

[0049] The named entity information obtaining module is used to obtain target text information and obtain specified named entity information in the target text information;

[0050] The attribute slot information obtaining module is used to determine the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, wherein the target attribute slot information is information for indicating the target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0051] The target attribute information obtaining module is used to extract attribute information of a first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine attribute information of a second target attribute according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information;

[0052] The text integration module is used to obtain the attribute information of the target attribute according to the attribute information of the first target attribute and the attribute information of the second target attribute;

[0053] The knowledge graph construction device is used to construct a knowledge graph for the target text information based on the attribute information of the target attribute and the specified named entity information.

[0054] In another aspect, the present application provides an electronic device, including:

[0055] Processor; and

[0056] The memory is used to store the program for the text information processing method. After the device is powered on and the program for the text information processing method is run by the processor, the following steps are performed:

[0057] Obtaining target text information, and obtaining specified named entity information in the target text information;

[0058] Determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0059] According to the specified named entity information and the target attribute slot information, attribute information of a first target attribute is extracted from the target text information, and attribute information of a second target attribute is determined according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information.

[0060] In another aspect of the present application, a storage device is provided, which stores a program for the text information processing method, and the program is executed by a processor to perform the following steps:

[0061] Obtaining target text information, and obtaining specified named entity information in the target text information;

[0062] Determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0063] According to the specified named entity information and the target attribute slot information, attribute information of a first target attribute is extracted from the target text information, and attribute information of a second target attribute is determined according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information.

[0064] Compared with the prior art, this application has the following advantages:

[0065] The text information processing method provided by the present application, after obtaining the target text information, first obtains the specified named entity information in the target text information, and then further determines the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, and finally, extracts the attribute information of the first target attribute that can directly determine the attribute information in the target text information from the target text information according to the specified named entity information and the target attribute slot information, and determines the attribute information of the preset second target attribute that cannot directly determine the attribute information in the target text information according to the target text information, the specified named entity information and the target attribute slot information. The text information processing method provided by the present application can not only extract the attribute information of the first target attribute that can directly determine the attribute information in the target text information from the target text information, but also can determine the attribute information of the preset second target attribute that cannot directly determine the attribute information in the target text information according to the target text information, the specified named entity information and the target attribute slot information, thereby improving the flexibility and accuracy of the extraction of "entity-attribute-attribute information" of the text information. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 This is a first schematic diagram of a scenario embodiment of the text information processing method provided in the present application.

[0067] Figure 2 This is a second schematic diagram of a scenario embodiment of the text information processing method provided in the present application.

[0068] Figure 3This is a third schematic diagram of a scenario embodiment of the text information processing method provided in the present application.

[0069] Figure 4 This is a fourth schematic diagram of a scenario embodiment of the text information processing method provided in the present application.

[0070] Figure 5 This is a flowchart of a text information processing method provided in the first embodiment of the present application.

[0071] Figure 6 This is a schematic diagram of a text information processing device provided in the second embodiment of the present application.

[0072] Figure 7 This is a schematic diagram of a text information processing system provided in the third embodiment of the present application.

[0073] Figure 8 This is a schematic diagram of a text information processing system provided in the fourth embodiment of the present application.

[0074] Fig. 9 A schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0075] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0076] In order to more clearly demonstrate the text information processing method provided by the present application, the application scenario of the text information processing method provided by the present application is first introduced. The text information processing method provided by the present application can be applied to the scenario where a client installed with a program or software for performing the text information processing method interacts with a server installed on the client that provides services for the program or software for performing the text information processing method, and can also be applied to the scenario where a client installed with a program or software for performing the text information processing method works alone, or to the scenario where the server works alone. Specifically, the client installed with a program or software for performing the text information processing method is an electronic device installed with a program or software for performing the text information processing method, such as a smart phone, a PC (Personal Computer), a tablet computer, etc. The server is a computer device that provides data, information provision, processing, and storage services for the program or software installed on the client, generally a server.

[0077] In the scenario embodiment of the text information processing method provided by the present application, the application scenario of the text information processing method provided by the present application is described in detail by taking the scenario of applying the text information processing method provided by the present application to a client installed with a program or software for performing the text information processing method and interacting with a server providing services for the program or software for performing the text information processing method installed on the client as an example. In the scenario embodiment of the text information processing method provided by the present application, text information refers to language information consisting of a sentence or a combination of multiple sentences with complete and systematic meanings, and text information refers to language information that can be composed of a sentence, paragraph language information consisting of a sentence or a combination of multiple sentences, or chapter language information. According to the scenario of generating text information and / or the purpose of text information, text information can be further divided into equipment quality text information and library management text information, etc. Specifically, power grid equipment quality text information includes but is not limited to: inspection logs, defect records, fault cases, disposal plans, technical specifications, standard guidelines, instruction manuals, instructions, scientific research papers, etc. Accordingly, in the scenario embodiment of the text information processing method provided by the present application, the target power grid equipment quality text information is the text information that needs to be processed to obtain the attribute information of the specified named entity in the text information. The present application specifically takes the target text information as the target power grid equipment quality text information as an example to describe in detail the scenario embodiment of the text information processing method provided by the present application. The specific process is as follows: Figure 1 , which is the first schematic diagram of a scenario implementation example of the text information processing method provided in this application.

[0078] The client 101 sends an information processing request message for the target power grid equipment quality text information to the server. Specifically, the client 101 first provides the target power grid equipment quality text information to the server 102, and then sends an information processing request message for the target power grid equipment quality text information to the server 102. Alternatively, the client 101 sends an information processing request message to the server 102, and the information processing request message carries the target power grid equipment quality text information. Other similar methods are also possible. After obtaining the information processing request message for the target power grid equipment quality text information sent by the client 101, the server 102 will first obtain the target power grid equipment quality text information, and then further obtain the specified named entity information in the target power grid equipment quality text information. The specific method is: perform named entity recognition on the target power grid equipment quality text information to obtain the specified named entity information.

[0079] In the scenario embodiment of the text information processing method provided in the present application, the named entity information is the information corresponding to the entity with special meaning or strong referentiality in the text information. As long as it is required by the business and is identified by a name, it can be called an entity, usually including names of people, places, organizations, time, proper nouns, etc. The designated named entity information is the named entity information pre-designated for specific text information. For example, the designated named entity information in the book management text information can be: the location and name of the library, the name and location of the bookshelf in the library, the status of the books on the bookshelf, etc. For example, the designated named entity information in the power grid equipment quality text information can be the name and location information of the power grid equipment, such as: West Lake Electric Well 1; component name information in the power grid equipment, such as: switch, on-load switch; fault phenomenon information, such as: short circuit, disconnection, etc. For details, please refer to Figure 2 , which is a second schematic diagram of a scenario embodiment of the text information processing method provided by the present application, Figure 2 The target power grid equipment quality text information is "The on-load switch of the West Lake First Transformer has a short circuit for one hour from 11:00 today", among which "West Lake First Transformer" corresponds to the name and location named entity information of the power grid equipment, "on-load switch" corresponds to the component name named entity information, and "short circuit" corresponds to the fault phenomenon named entity information.

[0080] Please refer to Figure 1 After obtaining the target power grid quality text information and the designated named entity information in the target power grid quality text information, the server 102 will further determine the target attribute slot information corresponding to the designated named entity information based on the designated named entity information. Among them, the target attribute slot information is information used to indicate the target attribute of the attribute information that needs to be filled in the slot corresponding to the designated named entity information. In the scenario embodiment of the text information processing method provided in the present application, the target attribute information includes the attribute field information corresponding to the designated named entity information and the value corresponding to the attribute field information, among which the attribute field information, such as: the type, size, type, status, start and end time of the equipment or component failure, etc., and the value corresponding to the attribute field information, such as: the on-load switch is a new type of on-load switch, the first short circuit is one hour, and the start and end time of the short circuit is 11 o'clock today, etc.

[0081] The so-called attribute slot is a position where one or a class of attributes can be inserted during the attribute extraction process of the specified named entity. The so-called attribute slot information is information used to identify the basic information of one or a class of attributes that need to be filled in the current slot. The so-called basic information of the attribute includes but is not limited to the type information of the target attribute, the name information of the target attribute, and the basic information of the target attribute. The so-called target attribute slot information is information used to indicate the target attribute that needs to fill in the slot corresponding to the specified named entity information, that is, the attribute slot pre-set according to the target attribute of the specified named entity, and the corresponding generated slot information.

[0082] Please refer to Figure 3 , which is the third schematic diagram of the scenario implementation example of the text information processing method provided by the present application. In the scenario implementation example of the text information processing method provided by the present application, the process of determining the target attribute slot information corresponding to the specified named entity information according to the specified named entity information is as follows: First, the type information corresponding to the specified named entity information is determined according to the specified named entity information. Then, the target entity named link corresponding to the power grid equipment quality text information is obtained. The target entity named link is a link pre-set for the power grid equipment quality text information and is used to obtain the target attribute slot information corresponding to the named entity information according to the type information corresponding to the named entity information. Finally, the target attribute slot information corresponding to the specified named entity information is determined according to the type information of the specified named entity and the target entity named link.

[0083] After the server 102 determines the target attribute slot information corresponding to the specified named entity information, it will extract the attribute information of the first target attribute from the target power grid equipment quality text information according to the specified named entity information and the target attribute slot information, and determine the attribute information of the second target attribute according to the target power grid equipment quality text information, the specified named entity information and the target attribute slot information. The first target attribute is a target attribute whose attribute information can be directly determined in the target power grid equipment quality text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target power grid equipment quality text information. The specific process includes the following two steps:

[0084] The first step is: first, input the target power grid equipment quality text information, the designated named entity information and the target attribute slot information into the first attribute information extraction module, and obtain the keyword information corresponding to the target power grid equipment quality text information, the entity feature information corresponding to the designated named entity information, the feature information corresponding to the target attribute and the position feature information of the designated named entity information in the target power grid equipment quality text information. Then, according to the keyword information corresponding to the target power grid equipment quality text information, the entity feature information corresponding to the designated named entity information, the feature information corresponding to the target attribute and the position feature information of the designated named entity information in the target power grid equipment quality text information, determine the attribute information of the first target attribute. Among them, before obtaining the attribute information of the first target attribute, the first target attribute will be obtained first. Among them, the first attribute information extraction module is used to determine the attribute information of the first target attribute according to the target power grid equipment quality text information, the designated named entity information and the target attribute slot information. For example, in "The on-load switch of the first transformer in West Lake has a short circuit for one hour starting from 11:00 today", the attribute information of the first target attribute includes: the attribute field of one target attribute of the named entity information of the fault phenomenon "short circuit" is "start and end time", and the corresponding value of this field is "starting from 11:00 today", and the attribute field of another target attribute of the named entity information of the fault phenomenon "short circuit" is "duration", and the corresponding value of this field is "one hour".

[0085] The specific implementation method of inputting the target text information and the designated named entity information into the first attribute information extraction module with the target attribute slot information to obtain the attribute information of the first target attribute is:

[0086] First, preprocess the target power grid equipment quality text information to obtain each word in the target power grid equipment quality text information.

[0087] Second, use the pre-trained word vector model to obtain the vectorized representation of each word in the target power grid equipment quality text information.

[0088] Third, the vectorized representation of each word in the target power grid equipment quality text information is input into the BI-LSTM (Bi-Directional Long-Short Term Memory) to obtain the vector matrix representation of the context information of the target power grid equipment quality text information. That is, the vectorized representation of each word in the target power grid equipment quality text information is input into a forward LSTM network to obtain the vector matrix representation of the context information of each word, and the vectorized representation of each word in the target power grid equipment quality text information is input into a reverse LSTM network to obtain the vector matrix representation of the context information of each word, and the vector matrix representation of the context information of each word and the vector matrix representation of the context information of each word are serially combined to obtain the vector matrix representation of the context information of the target power grid equipment quality text information.

[0089] Fourth, obtain the word vector representation of the pre-set words used to characterize the attributes, such as sensitive words that characterize abnormal state attributes, such as "exceeding the standard", "too high", "surge", etc., and words that characterize normal state attributes, such as "even", "stable", "good", etc.

[0090] Fifth, the encoding uses a pre-trained word vector model to encode the specified named entity and the first target attribute to obtain the word vector representation corresponding to the encoding of the specified named entity and the first target attribute. The vector matrix representation of the context information of the target power grid equipment quality text information, the pre-set word vector representation of the word used to represent the attribute, and the word vector representation corresponding to the encoding of the specified named entity and the first target attribute are merged through the attention model to obtain the first target attribute, and then the words used to represent the attribute in the target power grid equipment quality text information are classified through the softmax layer to obtain the attribute information of the first target attribute.

[0091] The second step is: first, according to the target attribute slot information and the first target attribute, determine the second target attribute, and determine the designated named entity corresponding to the second target attribute. Secondly, input the second target attribute, the designated named entity corresponding to the second target attribute, and the target power grid equipment quality text information into the second attribute information extraction module to obtain the feature information corresponding to the second target attribute. Thirdly, according to the feature information corresponding to the second target attribute, determine the candidate attribute information matching the feature information corresponding to the second target attribute in the target power grid equipment quality text information. Finally, according to the candidate attribute information, the second target attribute, the designated named entity corresponding to the second target attribute, and the target power grid equipment quality text information, obtain the attribute information of the second target attribute. Among them, the second attribute information extraction module is used to obtain the attribute information of the second target attribute according to the second target attribute, the designated named entity corresponding to the second target attribute, and the target power grid equipment quality text information. For example, in "The on-load switch of the first transformer of West Lake has a short circuit for one hour from 11 o'clock today", the attribute information of the second target attribute includes: the attribute field of a target attribute of the named entity information of the fault phenomenon "short circuit" is "occurrence state", and the value corresponding to this field is "occurrence".

[0092] The specific implementation method of inputting the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information into the second attribute information extraction module to obtain the attribute information of the second target attribute is:

[0093] First, preprocess the target power grid equipment quality text information to obtain each word in the target power grid equipment quality text information.

[0094] Second, the vectorized representation of each word in the target power grid equipment quality text information is input into the first BI-LSTM to obtain the vector matrix representation of the context information of the target power grid equipment quality text information. That is, the vectorized representation of each word in the target power grid equipment quality text information is input into a forward LSTM network to obtain the vector matrix representation of the context information of each word, and the vectorized representation of each word in the target power grid equipment quality text information is input into a reverse LSTM network to obtain the vector matrix representation of the context information of each word, and the vector matrix representation of the context information of each word and the vector matrix representation of the context information of each word are combined in series to obtain the vector matrix representation of the context information of the target power grid equipment quality text information. According to the vector matrix representation of the context information of the target power grid equipment quality text information, the vector representation annotation of the specified named entity in the target power grid equipment quality text information, the vector representation annotation of the target attribute, and the vector representation annotation of the position of the two keywords of the specified named entity and the target attribute in the target power grid equipment quality text information are obtained.

[0095] Third, the vectorized representation of each word in the target power grid equipment quality text information is input into the second BI-LSTM to obtain the vector matrix representation of the context information of the target power grid equipment quality text information. Using the vector matrix representation of the context information of the target power grid equipment quality text information, the vector representation of the target attribute in the target power grid equipment quality text information is obtained.

[0096] Fourth, the attention model is used to merge the annotations of the vector representation of the specified named entity in the target power grid equipment quality text information, the vector representation annotation of the target attribute, the vector representation annotation of the position of the two keywords of the specified named entity and the target attribute in the target power grid equipment quality text information, and the vector representation of the target attribute in the target power grid equipment quality text information, and then the conditional random field model is used to obtain the second target attribute and the attribute information of the second target attribute. It should be noted that the target attribute whose attribute information can be directly determined in the target text information and the preset target attribute whose attribute information cannot be directly determined in the target text information are pre-set based on the prophetic experience and attribute type, such as: the attribute with a specific value such as time and duration is the first target attribute, such as: the state attribute such as the occurrence state is the second target attribute.

[0097] It should be noted that in the scenario embodiment of the text information processing method provided by the present application, it is necessary to first obtain the first attribute information extraction module and the second attribute information extraction module through model training. Among them, the training process of the first attribute information extraction module is: first, determine the attribute information of the first target sample attribute; second, for the attribute information of the first target sample attribute, obtain the sample text information corresponding to the attribute information of the first target sample attribute, the specified named entity information in the sample text information, and the attribute slot information corresponding to the specified named entity information in the sample text information; third, according to the attribute information of the first target sample attribute, the sample text information corresponding to the attribute information of the first target sample attribute, the specified named entity information in the sample text information, and the attribute slot information corresponding to the specified named entity information in the sample text information, determine the first attribute information extraction module.

[0098] The training process of the second attribute information extraction module is as follows: first, determine the attribute information of the second target sample attribute; second, for the attribute information of the second target sample attribute, determine the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute; third, determine the second attribute information extraction module based on the attribute information of the second target sample attribute, the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute.

[0099] Please refer to Figure 1 After obtaining the attribute information of the first target attribute and the attribute information of the second target attribute, the scenario embodiment of the text information processing method provided by the present application also needs to obtain the attribute information of the target attribute based on the attribute information of the first target attribute and the attribute information of the second target attribute, and further determine the specified named entity information, target attribute and the set of attribute information of the target attribute corresponding to the target power grid equipment quality text information based on the attribute information of the first target attribute and the attribute information of the second target attribute, and provide it to the client 101. Among them, the process of obtaining the attribute information of the target attribute based on the attribute information of the first target attribute and the attribute information of the second target attribute is to deduplicate and merge the attribute information of the first target attribute and the attribute information of the second target attribute.

[0100] In summary, in the scenario embodiment of the text information processing method provided by the present application, the detailed process of the text information processing method is as follows: Figure 4As shown, it is the fourth schematic diagram of the scenario implementation method of the text information processing method provided by the present application. First, the target power grid equipment quality text information is obtained first; second, the target power grid equipment quality text information is subjected to named entity recognition to obtain the specified named entity information; third, according to the type information of the specified named entity and the target entity naming link, the target attribute slot information corresponding to the specified named entity information is determined; fourth, the target power grid equipment quality text information, the specified named entity information and the target attribute slot information are input into the first attribute information extraction module to obtain the attribute information of the first target attribute; and first, according to the target attribute slot information and the first target attribute, the second target attribute is determined, and the specified named entity corresponding to the second target attribute is determined; then the second target attribute, the specified named entity corresponding to the second target attribute and the target power grid equipment quality text information are input into the second attribute information extraction module to obtain c; fifth, the attribute information of the first target attribute and the attribute information of the second target attribute are merged to obtain the attribute information of the target attribute; sixth, according to the attribute information of the target attribute and the specified named entity information, the set of the specified named entity information, the target attribute and the attribute information of the target attribute corresponding to the target power grid equipment quality text information is determined.

[0101] The application scenarios of the text information processing method provided in the embodiments of the present application are not specifically limited. In addition to the above-mentioned application scenario embodiments, the text information processing method provided in the embodiments of the present application also has other application scenario embodiments, such as: the text information processing method provided in the embodiments of the present application can also be applied to other application scenarios outside the above-mentioned power industry. Specifically, the text information processing method provided in the embodiments can also be used in other application scenarios such as industrial manufacturing, mechanical processing, chemical manufacturing, and food processing.

[0102] When the text information processing method provided in the embodiment of the present application is applied to chemical manufacturing, the text information includes but is not limited to: operation and inspection logs of chemical equipment, instructions for use of chemical products, production technical specifications, chemical manufacturing standard guidelines, chemical manufacturing instruction manuals, etc. At this time, the named entity information is the information corresponding to entities with special meaning or strong referentiality in the text information, such as: chemical equipment, chemical products, operators, reaction time of chemical products, and product generation rate, etc.

[0103] When the text information processing method provided in the embodiments of the present application is applied to mechanical processing, the text information includes but is not limited to: the process of mechanical processing, the transportation and storage method of raw materials, the generation of technical specifications, mechanical processing standard guidelines, mechanical processing instruction manuals, etc. At this time, the named entity information is the information corresponding to the entity with special meaning or strong referentiality in the text information, such as: mechanical equipment, raw materials, operators, parts allocation time, and machine dimensions, etc. In the embodiments of the present application, the application of the text information processing method provided in the embodiments of the present application to other application scenarios other than the above-mentioned power industry will not be described in detail one by one.

[0104] The embodiments of the application scenarios of the above-mentioned text information processing method are provided to facilitate the understanding of the text information processing method provided by the present application, but are not used to limit the text information processing method provided by the present application.

[0105] First embodiment

[0106] In the first embodiment of the present application, a text information processing method is provided, such as Figure 5 As shown, it is a flowchart of a text information processing method provided in the first embodiment of the present application.

[0107] In step S501, target text information is obtained, and designated named entity information in the target text information is obtained.

[0108] In the first embodiment of the present application, text information refers to language information consisting of a sentence or a combination of multiple sentences with complete and systematic meanings, and text information refers to language information consisting of a sentence, paragraph language information consisting of a sentence or a combination of multiple sentences, or chapter language information. According to the scenario of generating text information and / or the purpose of text information, text information can be further divided into equipment quality text information and book management text information, etc. Specifically, power grid equipment quality text information includes but is not limited to: inspection logs, defect records, fault cases, disposal plans, technical specifications, standard guidelines, instruction manuals, instructions, scientific research papers, etc. Correspondingly, the target text information is text information that needs to be processed to obtain attribute information of specified named entities in the text information.

[0109] In the first embodiment of the present application, the designated named entity information in the target text information is obtained, and the specific method is: perform named entity recognition on the target text information to obtain the designated named entity information. Named entity information is information corresponding to entities with special meaning or strong referentiality in text information. As long as it is required by the business and a special text fragment with a name as an identifier can be called an entity, it usually includes names of people, places, institutions, time, proper nouns, etc. The designated named entity information is the named entity information pre-designated for specific text information. For example, the designated named entity information in the book management text information can be: the location and name of the library, the name and location of the bookshelf in the library, the status of the books on the bookshelf, etc. For example, the designated named entity information in the power grid equipment quality text information can be the name and location information of the power grid equipment, such as: West Lake Electric Well 1; component name information in the power grid equipment, such as: switch, load switch; fault phenomenon information, such as: short circuit, disconnection, etc.

[0110] In step S502, target attribute slot information corresponding to the specified named entity information is determined according to the specified named entity information, where the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information.

[0111] In the first embodiment of the present application, the target attribute information includes attribute field information corresponding to the specified named entity information and the value corresponding to the attribute field information, etc., wherein the attribute field information, such as when the target text information is the target power grid equipment quality text information: the type, size, type, status, start and end time of the power grid equipment or component, etc., and the value corresponding to the attribute field information, such as: the on-load switch is a new on-load switch, the first short circuit is one hour, the start and end time of the short circuit is 11 o'clock today, etc.

[0112] In the first embodiment of the present application, the method for determining the target attribute slot information corresponding to the specified named entity information is as follows: first, according to the specified named entity information, the type information corresponding to the specified named entity information is determined. Then, the target entity naming link corresponding to the target text information is obtained, and the target entity naming link is a link pre-set for the target text information and used to obtain the target attribute slot information corresponding to the named entity information according to the type information corresponding to the named entity information. Finally, according to the type information of the specified named entity and the target entity naming link, the target attribute slot information corresponding to the specified named entity information is determined.

[0113] In the first embodiment of the present application, the so-called attribute slot is a position where one or a class of attributes can be inserted during the attribute extraction process of the specified named entity. The so-called attribute slot information is information used to identify the basic information of one or a class of attributes that need to be filled in the current slot. The so-called basic information of the attribute includes but is not limited to the type information of the target attribute, the name information of the target attribute, and the basic information of the target attribute. The so-called target attribute slot information is information used to indicate the target attribute that needs to fill in the slot corresponding to the specified named entity information, that is, the attribute slot pre-set according to the target attribute of the specified named entity, and the corresponding generated slot information.

[0114] In step S503, attribute information of the first target attribute is extracted from the target text information according to the specified named entity information and the target attribute slot information, and attribute information of the second target attribute is determined according to the target text information, the specified named entity information and the target attribute slot information.

[0115] In the first embodiment of the present application, the so-called first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information. Specifically, the target attribute whose attribute information can be directly determined in the target text information and the preset target attribute whose attribute information cannot be directly determined in the target text information are pre-set based on the prophetic experience and the attribute type, such as: the attribute with a specific value such as time and duration is the first target attribute, such as: the state attribute such as the occurrence state is the second target attribute. In the first embodiment of the present application, the steps for determining the attribute information of the first target attribute and the attribute information of the second target attribute are as follows:

[0116] First, the target text information, the designated named entity information and the target attribute slot information are input into the first attribute information extraction module to obtain the attribute information of the first target attribute. The first attribute information extraction module is used to determine the attribute information of the first target attribute according to the target text information, the designated named entity information and the target attribute slot information. The target text information, the designated named entity information and the target attribute slot information are input into the first attribute information extraction module to obtain the attribute information of the first target attribute, including: inputting the target text information, the designated named entity information and the target attribute slot information into the first attribute information extraction module to obtain the first target attribute, and obtaining the attribute information of the first target attribute

[0117] Then, according to the target attribute slot information and the first target attribute, the second target attribute is determined, and the designated named entity corresponding to the second target attribute is determined.

[0118] Finally, the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into the second attribute information extraction module to obtain the attribute information of the second target attribute. The second attribute information extraction module is used to obtain the attribute information of the second target attribute based on the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information.

[0119] It should be noted that in the first embodiment of the present application, it is necessary to first obtain the first attribute information extraction module and the second attribute information extraction module through model training, and the specific process is as follows:

[0120] The process of obtaining the first attribute information extraction module is as follows: first, determining the attribute information of the first target sample attribute; second, for the attribute information of the first target sample attribute, obtaining the sample text information corresponding to the attribute information of the first target sample attribute, the specified named entity information in the sample text information, and the attribute slot information corresponding to the specified named entity information in the sample text information; third, determining the first attribute information extraction module based on the attribute information of the first target sample attribute, the sample text information corresponding to the attribute information of the first target sample attribute, the specified named entity information in the sample text information, and the attribute slot information corresponding to the specified named entity information in the sample text information.

[0121] The process of obtaining the second attribute information extraction module is as follows: first, determining the attribute information of the second target sample attribute; second, for the attribute information of the second target sample attribute, determining the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute; third, determining the second attribute information extraction module based on the attribute information of the second target sample attribute, the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute.

[0122] In the first embodiment of the present application, the target text information and the designated named entity information are input into the first attribute information extraction module with the target attribute slot information, and the steps of obtaining the attribute information of the first target attribute are as follows: first, the target text information and the designated named entity information are input into the first attribute information extraction module with the target attribute slot information, and the keyword information corresponding to the target text information, the entity feature information corresponding to the designated named entity information, the feature information corresponding to the target attribute, and the position feature information of the designated named entity information in the target text information are obtained. Then, the attribute information of the first target attribute is determined according to the keyword information corresponding to the target text information, the entity feature information corresponding to the designated named entity information, the feature information corresponding to the target attribute, and the position feature information of the designated named entity information in the target text information.

[0123] In the first embodiment of the present application, the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into the second attribute information extraction module, and the process of obtaining the attribute information of the second target attribute is as follows: first, the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into the second attribute information extraction module to obtain the feature information corresponding to the second target attribute. Then, according to the feature information corresponding to the second target attribute, the candidate attribute information matching the feature information corresponding to the second target attribute is determined in the target text information. Finally, according to the candidate attribute information, the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information, the attribute information of the second target attribute is obtained.

[0124] In the first embodiment of the present application, after obtaining the attribute information of the first target attribute and the attribute information of the second target attribute, it is also necessary to obtain the attribute information of the target attribute based on the attribute information of the first target attribute and the attribute information of the second target attribute, and further determine the set of the specified named entity information, the target attribute, and the attribute information of the target attribute corresponding to the target text information based on the attribute information of the first target attribute and the attribute information of the second target attribute. Among them, the process of obtaining the attribute information of the target attribute based on the attribute information of the first target attribute and the attribute information of the second target attribute is to deduplicate and merge the attribute information of the first target attribute and the attribute information of the second target attribute.

[0125] The text information processing method provided in the first embodiment of the present application, after obtaining the target text information, first obtains the specified named entity information in the target text information, and then further determines the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, and finally, extracts the attribute information of the first target attribute that can directly determine the attribute information in the target text information from the target text information according to the specified named entity information and the target attribute slot information, and determines the attribute information of the preset second target attribute that cannot directly determine the attribute information in the target text information according to the target text information, the specified named entity information and the target attribute slot information. The text information processing method provided in the first embodiment of the present application can not only extract the attribute information of the first target attribute that can directly determine the attribute information in the target text information from the target text information, but also can determine the attribute information of the preset second target attribute that cannot directly determine the attribute information in the target text information according to the target text information, the specified named entity information and the target attribute slot information, thereby improving the flexibility and accuracy of the extraction of "entity-attribute-attribute information" of the text information.

[0126] Second embodiment

[0127] Corresponding to the text information processing method provided in the application scenario embodiment and the first embodiment of the present application, the second embodiment of the present application provides a text information processing device. Since the device embodiment is basically similar to the application scenario embodiment and the first embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the application scenario embodiment and the first embodiment. The device embodiment described below is only illustrative.

[0128] Please refer to Figure 6 , which is a schematic diagram of a text information processing device provided in the second embodiment of the present application.

[0129] The text information processing device provided in the second embodiment of the present application includes:

[0130] A named entity information obtaining unit 601 is used to obtain target text information and obtain specified named entity information in the target text information;

[0131] The attribute slot information obtaining unit 602 is used to determine the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, wherein the target attribute slot information is information used to indicate the target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0132] The target attribute information obtaining unit 603 is used to extract the attribute information of the first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information.

[0133] Optionally, the first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information.

[0134] Optionally, the target attribute information obtaining unit 603 is specifically used to input the target text information, the specified named entity information and the target attribute slot information into a first attribute information extraction module to obtain the attribute information of the first target attribute. The first attribute information extraction module is used to determine the attribute information of the first target attribute based on the target text information, the specified named entity information and the target attribute slot information.

[0135] Optionally, the step of inputting the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information to obtain the attribute information of the first target attribute includes:

[0136] Input the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information, and obtain keyword information corresponding to the target text information, entity feature information corresponding to the designated named entity information, feature information corresponding to the target attribute, and position feature information of the designated named entity information in the target text information;

[0137] The attribute information of the first target attribute is determined according to the keyword information corresponding to the target text information, the entity feature information corresponding to the specified named entity information, the feature information corresponding to the target attribute and the position feature information of the specified named entity information in the target text information.

[0138] Optionally, also include:

[0139] Determining attribute information of a first target sample attribute;

[0140] For the attribute information of the first target sample attribute, obtain sample text information corresponding to the attribute information of the first target sample attribute, specified named entity information in the sample text information, and attribute slot information corresponding to the specified named entity information in the sample text information;

[0141] The first attribute information extraction module is determined according to the attribute information of the first target sample attribute, the sample text information corresponding to the attribute information of the first target sample attribute, the specified named entity information in the sample text information, and the attribute slot information corresponding to the specified named entity information in the sample text information.

[0142] Optionally, the step of inputting the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information to obtain the attribute information of the first target attribute comprises: inputting the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information to obtain the first target attribute, and obtaining the attribute information of the first target attribute;

[0143] The target attribute information obtaining unit 603 is used to determine the second target attribute and the designated named entity corresponding to the second target attribute according to the target attribute slot information and the first target attribute; the second target attribute, the designated named entity corresponding to the second target attribute and the target text information are input into a second attribute information extraction module to obtain the attribute information of the second target attribute; the second attribute information extraction module is used to obtain the attribute information of the second target attribute according to the second target attribute, the designated named entity corresponding to the second target attribute and the target text information.

[0144] Optionally, the step of inputting the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information into a second attribute information extraction module to obtain the attribute information of the second target attribute includes:

[0145] Inputting the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information into a second attribute information extraction module to obtain feature information corresponding to the second target attribute;

[0146] determining, in the target text information, candidate attribute information matching the feature information corresponding to the second target attribute according to the feature information corresponding to the second target attribute;

[0147] Attribute information of the second target attribute is obtained according to the candidate attribute information, the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information.

[0148] Optionally, also include:

[0149] Determining attribute information of a second target sample attribute;

[0150] For the attribute information of the second target sample attribute, determine the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute;

[0151] The second attribute information extraction module is determined according to the attribute information of the second target sample attribute, the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute.

[0152] Optionally, the named entity information obtaining unit 601 is specifically configured to perform named entity recognition on the target text information to obtain the specified named entity information.

[0153] Optionally, the attribute slot information obtaining unit is specifically used to determine type information corresponding to the specified named entity information based on the specified named entity information; obtain a target entity naming link corresponding to the target text information, the target entity naming link being a link pre-set for the target text information and used to obtain target attribute slot information corresponding to the named entity information based on the type information corresponding to the named entity information; determine the target attribute slot information corresponding to the specified named entity information based on the type information of the specified named entity and the target entity naming link.

[0154] Optionally, the text information processing device provided in the second embodiment of the present application further includes:

[0155] an attribute information integration unit, configured to obtain attribute information of the target attribute according to the attribute information of the first target attribute and the attribute information of the second target attribute;

[0156] The target text information set obtaining unit is used to determine the specified named entity information corresponding to the target text information, the target attribute and the set of the attribute information of the target attribute according to the attribute information of the target attribute and the specified named entity information.

[0157] Third embodiment

[0158] Corresponding to the text information processing method provided by the application scenario embodiment and the first embodiment of the present application, the third embodiment of the present application provides a text information processing system. Since the third embodiment is basically similar to the application scenario embodiment and the first embodiment, the description is relatively simple, and the relevant parts refer to the partial description of the application scenario embodiment and the first embodiment. The third embodiment described below is only illustrative.

[0159] Please refer to Figure 7 , which is a schematic diagram of a text information processing system provided in the third embodiment of the present application.

[0160] The text information processing system provided in the third embodiment of the present application includes: a text collection device 701, a text processing device 702 and a text integration device 703, wherein the text processing device 702 includes: a named entity information acquisition module 702-1, an attribute slot information acquisition module 702-2 and a target attribute information acquisition module 702-3;

[0161] The text collection device 701 is used to collect target text information;

[0162] The named entity information obtaining module 702-1 is used to obtain target text information and obtain designated named entity information in the target text information;

[0163] The attribute slot information obtaining module 702-2 is used to determine the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, wherein the target attribute slot information is information indicating the target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0164] The target attribute information obtaining module 702-3 is used to extract the attribute information of the first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information;

[0165] The so-called first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information;

[0166] The text integration device 703 is used to obtain the attribute information of the target attribute based on the attribute information of the first target attribute and the attribute information of the second target attribute, and determine the specified named entity information, the target attribute and the set of the attribute information of the target attribute corresponding to the target text information based on the attribute information of the target attribute and the specified named entity information.

[0167] It should be noted that for the detailed description of the text information processing system provided in the third embodiment of the present application, reference can be made to the relevant description of the application scenario embodiment of the present application and the text information processing method provided in the first embodiment, which will not be repeated here.

[0168] Fourth embodiment

[0169] Corresponding to the text information processing method provided in the application scenario embodiment and the first embodiment of the present application, the fourth embodiment of the present application provides a knowledge graph construction system. Since the fourth embodiment is basically similar to the application scenario embodiment and the first embodiment, the description is relatively simple, and the relevant parts refer to the application scenario embodiment and the partial description of the first embodiment. The fourth embodiment described below is only exemplary.

[0170] Please refer to Figure 8 , which is a schematic diagram of a text information processing system provided in the fourth embodiment of the present application.

[0171] The text information processing system provided in the fourth embodiment of the present application includes: a text processing device 801 and a knowledge graph construction device 802, wherein the text processing device 801 includes: a text collection module 801-1, a named entity information acquisition module 801-2, an attribute slot information acquisition module 801-3, a target attribute information acquisition module 801-4 and a text integration module 801-5;

[0172] The text collection module 801-1 is used to collect target text information;

[0173] The named entity information obtaining module 801-2 is used to obtain target text information and obtain designated named entity information in the target text information;

[0174] The attribute slot information obtaining module 801-3 is used to determine the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, wherein the target attribute slot information is information indicating the target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0175] The target attribute information obtaining module 801-4 is used to extract the attribute information of the first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information;

[0176] The so-called first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information;

[0177] The text integration module 801-5 is used to obtain the attribute information of the target attribute according to the attribute information of the first target attribute and the attribute information of the second target attribute;

[0178] The knowledge graph construction device 802 is used to construct a knowledge graph for the target text information based on the attribute information of the target attribute and the specified named entity information.

[0179] It should be noted that for the detailed description of the text information processing system provided in the fourth embodiment of the present application, reference can be made to the relevant description of the application scenario embodiment of the present application and the text information processing method provided in the first embodiment, which will not be repeated here.

[0180] Fifth embodiment

[0181] Corresponding to the application scenario embodiment of the present application and the text information processing method provided in the first embodiment, the fifth embodiment of the present application provides an electronic device.

[0182] like Fig. 9 As shown, it is a schematic diagram of an electronic device provided in an embodiment of the present application.

[0183] The electronic device comprises:

[0184] Processor 901; and

[0185] The memory 902 stores a computer program. After the device is powered on and the processor runs the computer program, the following steps are performed:

[0186] Obtaining target text information, and obtaining specified named entity information in the target text information;

[0187] Determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0188] According to the designated named entity information and the target attribute slot information, attribute information of a first target attribute is extracted from the target text information, and according to the target text information, the designated named entity information and the target attribute slot information, attribute information of a second target attribute is determined.

[0189] The so-called first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information.

[0190] It should be noted that for the detailed description of the electronic device provided in the fifth embodiment of the present application, reference can be made to the relevant description of the application scenario embodiment of the present application and the text information processing method provided in the first embodiment, which will not be repeated here.

[0191] Sixth embodiment

[0192] Corresponding to the data processing method provided in the above embodiments of the present application, the sixth embodiment of the present application provides a storage medium, the storage medium stores a computer program, the computer program is executed by a processor, and executes the following steps:

[0193] Obtaining target text information, and obtaining specified named entity information in the target text information;

[0194] Determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information;

[0195] According to the designated named entity information and the target attribute slot information, attribute information of a first target attribute is extracted from the target text information, and according to the target text information, the designated named entity information and the target attribute slot information, attribute information of a second target attribute is determined.

[0196] The so-called first target attribute is a preset target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is a preset target attribute whose attribute information cannot be directly determined in the target text information.

[0197] It should be noted that for the detailed description of the storage medium provided in the sixth embodiment of the present application, reference can be made to the relevant description of the application scenario embodiment of the present application and the text information processing method provided in the first embodiment, which will not be repeated here.

[0198] Although the present application is disclosed as above in the preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

[0199] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0200] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash memory (Flash RAM). The memory is an example of a computer-readable medium.

[0201] 1. Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined in this article, computer-readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0202] 2. Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

Claims

1. A text information processing method, characterized in that: include: Obtaining target text information, and obtaining specified named entity information in the target text information; Determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information; Extracting attribute information of a first target attribute from the target text information according to the designated named entity information and the target attribute slot information, and determining attribute information of a second target attribute according to the target text information, the designated named entity information and the target attribute slot information, wherein the first target attribute is the target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is the target attribute whose attribute information cannot be directly determined in the target text information; The step of determining target attribute slot information corresponding to the specified named entity information according to the specified named entity information includes: According to the specified named entity information, determine the type information corresponding to the specified named entity information; obtain the target entity naming link corresponding to the target text information, the target entity naming link being a link pre-set for the target text information and used to obtain the target attribute slot information corresponding to the named entity information according to the type information corresponding to the named entity information; determine the target attribute slot information corresponding to the specified named entity information according to the type information of the specified named entity and the target entity naming link; The determining the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information includes: According to the target attribute slot information and the first target attribute, the second target attribute is determined, and the designated named entity corresponding to the second target attribute is determined; the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into a second attribute information extraction module to obtain the attribute information of the second target attribute.

2. The text information processing method according to claim 1, characterized in that: The step of extracting the attribute information of the first target attribute from the target text information according to the specified named entity information and the target attribute slot information comprises: inputting the target text information, the specified named entity information and the target attribute slot information into a first attribute information extraction module to obtain the attribute information of the first target attribute, wherein the first attribute information extraction module is used to determine the attribute information of the first target attribute according to the target text information, the specified named entity information and the target attribute slot information.

3. The text information processing method according to claim 2, characterized in that: The step of inputting the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information to obtain the attribute information of the first target attribute includes: Input the target text information and the designated named entity information into a first attribute information extraction module with the target attribute slot information, and obtain keyword information corresponding to the target text information, entity feature information corresponding to the designated named entity information, feature information corresponding to the target attribute, and position feature information of the designated named entity information in the target text information; The attribute information of the first target attribute is determined according to the keyword information corresponding to the target text information, the entity feature information corresponding to the specified named entity information, the feature information corresponding to the target attribute and the position feature information of the specified named entity information in the target text information.

4. The text information processing method according to claim 2, characterized in that: Also includes: Determining attribute information of a first target sample attribute; For the attribute information of the first target sample attribute, obtain sample text information corresponding to the attribute information of the first target sample attribute, specified named entity information in the sample text information, and attribute slot information corresponding to the specified named entity information in the sample text information; The first attribute information extraction module is determined according to the attribute information of the first target sample attribute, the sample text information corresponding to the attribute information of the first target sample attribute, the specified named entity information in the sample text information, and the attribute slot information corresponding to the specified named entity information in the sample text information.

5. The text information processing method according to claim 2, characterized in that: The step of inputting the target text information and the designated named entity information into the first attribute information extraction module with the target attribute slot information to obtain the attribute information of the first target attribute includes: inputting the target text information and the designated named entity information into the first attribute information extraction module with the target attribute slot information to obtain the first target attribute, and obtaining the attribute information of the first target attribute.

6. The text information processing method according to claim 5, characterized in that: The step of inputting the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information into a second attribute information extraction module to obtain the attribute information of the second target attribute includes: Inputting the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information into a second attribute information extraction module to obtain feature information corresponding to the second target attribute; determining, in the target text information, candidate attribute information matching the feature information corresponding to the second target attribute according to the feature information corresponding to the second target attribute; Attribute information of the second target attribute is obtained according to the candidate attribute information, the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information.

7. The text information processing method according to claim 6, characterized in that: Also includes: Determining attribute information of a second target sample attribute; For the attribute information of the second target sample attribute, determine the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute; A second attribute information extraction module is determined according to the attribute information of the second target sample attribute, the named entity information corresponding to the attribute information of the second target sample attribute, the attribute slot information corresponding to the attribute information of the second target sample attribute, and the text information corresponding to the attribute information of the second target sample attribute.

8. The text information processing method according to claim 1, characterized in that: The obtaining of target text information and obtaining designated named entity information in the target text information includes: performing named entity recognition on the target text information to obtain the designated named entity information.

9. The text information processing method according to claim 1, characterized in that: Also includes: Obtaining attribute information of the target attribute according to the attribute information of the first target attribute and the attribute information of the second target attribute; According to the attribute information of the target attribute and the designated named entity information, a set of the designated named entity information, the target attribute, and the attribute information of the target attribute corresponding to the target text information is determined.

10. A text information processing device, characterized in that: include: A named entity information obtaining unit, used to obtain target text information and obtain designated named entity information in the target text information; An attribute slot information obtaining unit, configured to determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information indicating a target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information; The step of determining target attribute slot information corresponding to the specified named entity information according to the specified named entity information includes: According to the specified named entity information, determine the type information corresponding to the specified named entity information; obtain the target entity naming link corresponding to the target text information, the target entity naming link being a link pre-set for the target text information and used to obtain the target attribute slot information corresponding to the named entity information according to the type information corresponding to the named entity information; determine the target attribute slot information corresponding to the specified named entity information according to the type information of the specified named entity and the target entity naming link; a target attribute information obtaining unit, configured to extract attribute information of a first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine attribute information of a second target attribute according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is the target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is the target attribute whose attribute information cannot be directly determined in the target text information; The determining the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information includes: According to the target attribute slot information and the first target attribute, the second target attribute is determined, and the designated named entity corresponding to the second target attribute is determined; the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into a second attribute information extraction module to obtain the attribute information of the second target attribute.

11. A text information processing system, characterized in that: include: A text collection device, a text processing device and a text integration device, wherein the text processing device comprises: a named entity information acquisition module, an attribute slot information acquisition module and a target attribute information acquisition module; The text collection device is used to collect target text information; The named entity information obtaining module is used to obtain target text information and obtain specified named entity information in the target text information; The attribute slot information obtaining module is used to determine the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, wherein the target attribute slot information is information for indicating the target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information; The step of determining target attribute slot information corresponding to the specified named entity information according to the specified named entity information includes: According to the specified named entity information, determine the type information corresponding to the specified named entity information; obtain the target entity naming link corresponding to the target text information, the target entity naming link being a link pre-set for the target text information and used to obtain the target attribute slot information corresponding to the named entity information according to the type information corresponding to the named entity information; determine the target attribute slot information corresponding to the specified named entity information according to the type information of the specified named entity and the target entity naming link; The target attribute information obtaining module is used to extract attribute information of a first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine attribute information of a second target attribute according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is the target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is the target attribute whose attribute information cannot be directly determined in the target text information; The text integration device is used to obtain the attribute information of the target attribute according to the attribute information of the first target attribute and the attribute information of the second target attribute, and determine the specified named entity information corresponding to the target text information, the target attribute, and the set of the attribute information of the target attribute according to the attribute information of the target attribute and the specified named entity information; The determining the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information includes: According to the target attribute slot information and the first target attribute, the second target attribute is determined, and the designated named entity corresponding to the second target attribute is determined; the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into a second attribute information extraction module to obtain the attribute information of the second target attribute.

12. A knowledge graph construction system, characterized in that: include: A text processing device and a knowledge graph construction device, wherein the text processing device comprises: a text collection module, a named entity information acquisition module, an attribute slot information acquisition module, a target attribute information acquisition module and a text integration module; The text collection module is used to collect target text information; The named entity information obtaining module is used to obtain target text information and obtain specified named entity information in the target text information; The attribute slot information obtaining module is used to determine the target attribute slot information corresponding to the specified named entity information according to the specified named entity information, wherein the target attribute slot information is information for indicating the target attribute of the attribute information that needs to be filled in the slot corresponding to the specified named entity information; The step of determining target attribute slot information corresponding to the specified named entity information according to the specified named entity information includes: According to the specified named entity information, determine the type information corresponding to the specified named entity information; obtain the target entity naming link corresponding to the target text information, the target entity naming link being a link pre-set for the target text information and used to obtain the target attribute slot information corresponding to the named entity information according to the type information corresponding to the named entity information; determine the target attribute slot information corresponding to the specified named entity information according to the type information of the specified named entity and the target entity naming link; The target attribute information obtaining module is used to extract attribute information of a first target attribute from the target text information according to the specified named entity information and the target attribute slot information, and determine attribute information of a second target attribute according to the target text information, the specified named entity information and the target attribute slot information, wherein the first target attribute is the target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is the target attribute whose attribute information cannot be directly determined in the target text information; The determining the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information includes: Determine the second target attribute according to the target attribute slot information and the first target attribute, and determine the designated named entity corresponding to the second target attribute; input the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information into a second attribute information extraction module to obtain attribute information of the second target attribute; The text integration module is used to obtain the attribute information of the target attribute according to the attribute information of the first target attribute and the attribute information of the second target attribute; The knowledge graph construction device is used to construct a knowledge graph for the target text information based on the attribute information of the target attribute and the specified named entity information.

13. An electronic device, characterized in that: include: processor; as well as The memory is used to store a program for a text information processing method. After the device is powered on and the program for the text information processing method is run by the processor, the following steps are performed: Obtaining target text information, and obtaining specified named entity information in the target text information; Determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information; Extracting attribute information of a first target attribute from the target text information according to the designated named entity information and the target attribute slot information, and determining attribute information of a second target attribute according to the target text information, the designated named entity information and the target attribute slot information, wherein the first target attribute is the target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is the target attribute whose attribute information cannot be directly determined in the target text information; The step of determining target attribute slot information corresponding to the specified named entity information according to the specified named entity information includes: According to the specified named entity information, determine the type information corresponding to the specified named entity information; obtain the target entity naming link corresponding to the target text information, the target entity naming link being a link pre-set for the target text information and used to obtain the target attribute slot information corresponding to the named entity information according to the type information corresponding to the named entity information; determine the target attribute slot information corresponding to the specified named entity information according to the type information of the specified named entity and the target entity naming link; The determining the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information includes: According to the target attribute slot information and the first target attribute, the second target attribute is determined, and the designated named entity corresponding to the second target attribute is determined; the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into a second attribute information extraction module to obtain the attribute information of the second target attribute.

14. A storage device, characterized in that: A program for a text information processing method is stored, and the program is run by a processor to perform the following steps: Obtaining target text information, and obtaining specified named entity information in the target text information; Determine, according to the specified named entity information, target attribute slot information corresponding to the specified named entity information, wherein the target attribute slot information is information for indicating a target attribute of attribute information that needs to be filled in the slot corresponding to the specified named entity information; Extracting attribute information of a first target attribute from the target text information according to the designated named entity information and the target attribute slot information, and determining attribute information of a second target attribute according to the target text information, the designated named entity information and the target attribute slot information, wherein the first target attribute is the target attribute whose attribute information can be directly determined in the target text information, and the second target attribute is the target attribute whose attribute information cannot be directly determined in the target text information; The step of determining target attribute slot information corresponding to the specified named entity information according to the specified named entity information includes: According to the specified named entity information, determine the type information corresponding to the specified named entity information; obtain the target entity naming link corresponding to the target text information, the target entity naming link being a link pre-set for the target text information and used to obtain the target attribute slot information corresponding to the named entity information according to the type information corresponding to the named entity information; determine the target attribute slot information corresponding to the specified named entity information according to the type information of the specified named entity and the target entity naming link; The determining the attribute information of the second target attribute according to the target text information, the specified named entity information and the target attribute slot information includes: According to the target attribute slot information and the first target attribute, the second target attribute is determined, and the designated named entity corresponding to the second target attribute is determined; the second target attribute, the designated named entity corresponding to the second target attribute, and the target text information are input into a second attribute information extraction module to obtain the attribute information of the second target attribute.

Images