An information collaborative processing platform and information query method based on a knowledge graph

Abstract

The application discloses a kind of information collaborative processing platform and information query method based on knowledge graph, including data acquisition processing layer, knowledge extraction intelligent analysis layer and knowledge application decision layer;The data acquisition processing layer includes data acquisition module, data format conversion module and knowledge base management module;The knowledge extraction intelligent analysis layer includes keyword extraction module, keyword hierarchical module, semantic fusion module and knowledge ontology dynamic evolution module;The knowledge application decision layer includes database module, entity and relationship extraction module, event topic library, organization library, evaluation result acquisition module and military strategy guidance information update module and military strategy graph generation module.The application can realize the effective cooperation between military strategy display, military strategy dynamic adjustment and intelligence data real-time processing, and can also provide more perfect military strategy related information query result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information data processing, and particularly relates to an information collaborative processing platform based on a knowledge graph and an information query method. BACKGROUND

[0002] With the progress of information technology, existing battlefield intelligence information presents the characteristics of explosive growth, complex sources, and diverse forms of representation. Traditional search engines and query methods cannot meet the capture and understanding of battlefield information and a large amount of military knowledge.

[0003] The application with the publication number CN115878812A discloses a strategic intent analysis method based on open source intelligence, which includes obtaining open source intelligence text, preprocessing; extracting multiple event elements, and extracting the core words of the event; extracting event dependency subgraph containing meta-event pairs; obtaining the annotation label corresponding to each meta-event in the meta-event pair; constructing a preliminary affairs graph; obtaining an optimized affairs graph; predicting subsequent events of the events in the optimized affairs graph through a strategic event prediction model, and obtaining the inducement of the events in the optimized affairs graph. The method provided in the application mines information from unstructured text, integrates related data and resources, and expands the open source intelligence text dataset. The event elements and event relationships in the open source intelligence text dataset are extracted. The enemy's strategic intent is analyzed through the affairs graph, realizing the transformation of traditional intelligence work to informationization and intelligentization, and providing auxiliary decision-making ability for intelligence analysis work.

[0004] The application with the publication number CN115878811A discloses a military intelligence intelligent analysis and deduction method based on an affairs graph. The application proposes an idea of intelligence analysis and deduction centered on events, from the automatic construction of an affairs graph to the visual reasoning centered on events, and deeply describes the specific process. The application uses event-driven intelligence analysis technology to comprehensively monitor and visually analyze battlefield intelligence data, which can provide a visual, manageable, and controllable military command decision-making platform for managers, help users to mine the value of military intelligence data, and improve the command decision-making efficiency of combat personnel.

[0005] The invention with publication number CN114896387A discloses a military intelligence analysis visualization method, device and computer readable storage medium. After obtaining and preprocessing event texts of various fields, the event classes of the event texts are divided and the event elements of the event texts are extracted according to an event ontology library. The main content and event class relationship of the event texts are extracted according to the event elements. The event evolution path is predicted and deduced through a Bayesian network classification model, and the event class relationship and the event evolution path are visualized. Thus, the overall context of the intelligence event can be displayed in the form of a graph as much as possible, and the related impact of the event can be predicted through a machine learning classification model, and customized field event content is pushed to the corresponding user.

[0006] However, the aforementioned invention needs to construct a matter reason map, optimize on the basis of the matter reason map, analyze the strategic intention of the enemy, and can only provide part of the information as a reference for military strategy. The information provided depends on the construction of the matter reason map, ignores the real-time changes of military strategy, cannot realize effective cooperation among military strategy display, military strategy dynamic adjustment and real-time processing of intelligence data, and cannot provide a relatively complete military intelligence information query function.

[0007] The invention with publication number CN115408532A discloses a weapon equipment knowledge graph construction method for open source intelligence. Military text data is obtained from open source resources, the military text data is preprocessed to obtain standardized military text data, the standardized military text data is labeled to obtain a training set (weapon equipment entity recognition data set to be trained, weapon equipment attribute extraction data set to be trained, and weapon equipment entity linking data set to be trained), the model is trained to improve the model, and the improved model is used to identify and attribute extract weapons and equipment. Finally, a knowledge graph is constructed, solving the problems of low utilization rate of military open source information, time-consuming and laborious query, and inconvenience caused by scattered, uneven quality, and large amount of data of current open source military information. The invention can only be used to construct a weapon equipment knowledge graph, cannot realize effective cooperation among military strategy display, military strategy dynamic adjustment and real-time processing of intelligence data, and can only query specified weapon state information, and cannot provide a relatively complete military intelligence information query function. SUMMARY

[0008] The military intelligence data convergence collaborative processing platform guided by military strategy is provided, which is guided by real-time military strategy and researches the military knowledge graph in the field of military intelligence information based on actual military field business data, integrates a large number of scattered and isolated intelligence, realizes the correlation of military intelligence information from the semantic level, provides strong support for the comprehensive analysis, research and display of intelligence information, improves the analysis efficiency, realizes the effective cooperation among military strategy display, military strategy dynamic adjustment and real-time processing of intelligence data, and can provide strong and complete military intelligence information query function.

[0009] To achieve the above technical purposes, the technical scheme adopted by the present application is:

[0010] An information collaborative processing platform based on a knowledge graph, comprising a data acquisition and processing layer, a knowledge extraction and intelligent analysis layer and a knowledge application and decision-making layer.

[0011] The data acquisition and processing layer comprises a data acquisition module, a data format conversion module and a knowledge base management module.

[0012] The data acquisition module obtains military intelligence field business data from multiple data sources, stores the data into corresponding databases after cleaning and classification, and the military intelligence field business data comprises various types of military intelligence achievement data analyzed and researched in recent years and intelligence information reported by departments at all levels in real time; the data format conversion module is used for uniformly representing the multi-source heterogeneous military intelligence field business data collected by the data acquisition module; the knowledge base management module is used for generating a semantic network knowledge base according to the military intelligence field business data after format conversion, dynamically expanding the semantic network knowledge base, and fusing repeated content and conflict content in the construction of the knowledge ontology; the semantic network knowledge base comprises four types of field knowledge ontology of person, organization, resource and military scene event, each field knowledge entity has corresponding attribute information, and each field knowledge entity comprises an active state attribute.

[0013] The knowledge extraction and intelligent analysis layer comprises a keyword extraction module, a keyword hierarchical module, a semantic fusion module and a knowledge ontology dynamic evolution module.

[0014] The keyword extraction module is used for extracting a plurality of words or phrases with greater relevance to entities as basic keywords.

[0015] The keyword hierarchical module constructs a core keyword set of the current period based on the evolution trend of historical military strategy related information, the military strategy guidance information of the current period and the information query data of military experts; then, the remaining basic keywords are hierarchically and classified according to the dependency relationship of syntactic analysis and the keyword co-occurrence frequency, and a spatial distribution model between keywords and entities and between keywords and keywords is constructed.

[0016] The semantic fusion module takes the association between the keywords and the knowledge ontology of each field as the main feature, establishes the feature vector of the knowledge ontology of each field, and calculates the similarity between the knowledge ontology of each field. According to the general threshold, the relationship between the knowledge ontology of each field is obtained; starting from the keyword frequency and distance features, the hierarchy and sequence relationship between the knowledge entities of each field are mined, the knowledge ontology evolution semantic network is constructed, and the multi-source heterogeneous data in the semantic network knowledge base is fused;

[0017] The knowledge ontology dynamic evolution module starts from the dynamic evolution of each field knowledge ontology itself and each other, performs upper and lower semantic reasoning on the fused semantic network knowledge base, extends the link after finding the link between the two knowledge ontologies, discovers the potential semantics or more extensive relationship on the link, mines the evolution trend of the multi-element relationship between the person, organization, resource behavior sequence and military scene event, and constructs a multi-element knowledge semantic network. The network topology is decoupled and accumulated by introducing the time slice mode, and the position label of each node and link edge is set in each time slice to indicate whether the network node and edge exist in the corresponding time slice. The state transition is used to discover and reason the knowledge group, and the active state attribute of each field knowledge entity is updated;

[0018] The knowledge application decision layer includes a database module, an entity and relationship extraction module, an event topic library, an organization library, an evaluation result collection module, a military strategy guidance information update module and a military strategy graph generation module.

[0019] The database module is used to store the data classified and cleaned by the data collection module, and add message attributes to each message in the database. The entity and relationship extraction module is used to extract and display entity attribute information, entity graph information and entity relationship graph according to the input entity keywords. The event topic library is used to manage the related information of the events in the active state in the current period. The organization library is used to manage the related information of all organizations. The military strategy graph generation module is used to determine the evolution state and evolution trend of the military strategy related information from the aspects of person, organization, resource and military scene event, and update the active state attribute of each field knowledge entity. According to the updated active state attribute, a number of key attention persons, key attention organizations, key attention resources and key attention events in the current period are selected to generate a military strategy graph. The evaluation result collection module collects the evaluation results of military experts on military strategy related information. The military strategy guidance information update module dynamically updates the military strategy guidance information of the next period combined with the evolution state and evaluation results of the military strategy related information.

[0020] Further, the database module includes a database management component, a database query component and a database display component; the database management component is used to store the data classified by the data collection module into the two types of databases, i.e. the basic database and the trend database, and add the time of entering the database, the date of issuing the message and the key word attribute to each message entering the database; the database query component provides the data query function, including the date range of issuing the message plug-in, the key word query plug-in and the database type, searches the message in the selected database through the date range of issuing the message and the key word input by the user, and calls the database display component to push the search result to the display large screen.

[0021] Further, the entity and relationship extraction module includes an entity recognition component, an entity graph component and a relationship recognition component; the entity recognition component queries the related entity according to the input key word and pushes the related attribute of the query entity to the display large screen; the entity graph component includes a first entity name plug-in, the entity graph component queries the related node of the entity in the knowledge graph according to the input single entity name, obtains and displays the associated attribute and relationship query result; the relationship recognition component includes a relationship selection plug-in and a plurality of second entity name plug-ins, the relationship recognition component inquires and displays the knowledge graph of the specified relationship between the input multiple entities through the multi-layer penetration between the multiple entities according to the input multiple relationship types; the specified relationship includes the association relationship, the associated organization and the associated event between the entities.

[0022] Further, the event special topic database includes an event analysis component, an event adding component, an event association component, an associated message aggregation component and an associated message query component; the event analysis component analyzes each key event from multiple dimensions including the event basic information, the event introduction, the development history, the associated message, the relationship graph and the event hot words; the event adding component is used to manually add the key event; the event association component is used to summarize and analyze different event topics generated by the development and evolution of the event; the associated message aggregation component is used to aggregate the messages related to the event topic according to the time axis; and the associated message query component is used to provide the associated message viewing function.

[0023] Further, the organization library comprises an organization basic information management component, an organization experience management component, an associated event management component, an associated message management component and a relationship graph generation component; the organization basic information management component is configured to manage the basic information of all organization entities in the organization library; the organization experience management component is configured to manage the relevant person structure and resource information change process of the organization; the associated event management component is configured to add event types related to all organization entities in real time according to the knowledge reasoning result; the associated message management component is configured to manage all messages related to the organization; and the relationship graph generation component is configured to sort and generate a binary relationship knowledge graph related to the organization.

[0024] Further, the knowledge base management module comprises an entity construction component, a semantic network knowledge base generation component and a conflict fusion component.

[0025] The entity construction component is configured to study the characteristics of part of unstructured text, manually construct an initial military field knowledge entity attribute library, expand the military field knowledge entity attribute library based on word similarity, and label entities and attributes in the unstructured text by using a dictionary method combined with a part-of-speech attribute determination method.

[0026] The semantic network knowledge base generation component takes the spatial position co-occurrence characteristics of keywords as a training set, trains the labeled entity and attribute data by using a deep learning model, extracts new field knowledge data by using the trained deep learning model, evaluates the effectiveness of the extracted entities and attributes, dynamically expands the evaluated effective entities and attributes to the military field knowledge entity attribute library, and identifies field knowledge entities and attributes in a large amount of text based on a reinforcement machine learning method.

[0027] The conflict fusion component analyzes term conflicts, semantic conflicts and predicate conflicts, and adopts a hybrid method of logic tree fusion, frequency fusion and syntax fusion to select and fuse repeated content and conflicting content in the construction of the knowledge ontology.

[0028] Further, the conflict fusion component comprises a term library, a predicate library, an ontology library, a term conflict processing unit, a semantic conflict processing unit, a predicate conflict processing unit and a fusion unit.

[0029] The term library and the predicate library respectively extract relevant knowledge ontology from the military field knowledge entity attribute library, respectively construct a term set, a predicate set and a semantic set and transmit to the ontology library; the ontology library analyzes the term set, the predicate set and the semantic set, and obtains existing term conflicts, predicate conflicts and semantic conflicts; the term conflict processing unit calls a logical tree fusion model, a frequency fusion model and a syntax fusion model to process the term conflicts, the semantic conflict processing unit calls the syntax fusion model to process the semantic conflicts, and the predicate conflict processing unit calls the frequency fusion model and the syntax fusion model to process the predicates; the fusion unit comprehensively processes the processing results of the term conflict processing unit, the semantic conflict processing unit and the predicate conflict processing unit, and generates a fused knowledge item;

[0030] The logical tree fusion model fuses the conflict items by judging the logical tree relationship of the two conflict items; the frequency fusion model uses high-frequency co-occurring knowledge items as the highest score output; and the syntax fusion model uses a syntax dependency tree discriminant to judge the semantic space dependency relationship between entity, attribute words and attribute value words, and selects multiple items that produce conflicts.

[0031] Further, the keyword hierarchical module filters the entities in an active state in the current period based on the evolution trend of historical military strategy related information, and selects a first keyword related to the entity from the basic keywords according to the association relationship between each entity and the keyword; obtains and analyzes the military strategy guidance information of the current period, and selects a second keyword related to the entity from the basic keywords according to the entity corresponding to the military strategy guidance information; and comprehensively selects the first keyword and the second keyword to form a core keyword set of the current period.

[0032] Further, the knowledge ontology dynamic evolution module includes a unit knowledge reasoning component, a multi-element relationship reasoning component and a dynamic knowledge reasoning component.

[0033] The unit knowledge reasoning component loads knowledge network data, establishes a data structure of neighbor nodes and label information corresponding to each node, calculates the probability value of the neighbor nodes of each node in the knowledge network being walked to according to the label attributes of the node and its neighbor nodes and the specified label information proportion adjustable parameter p, and randomly selects several times from the neighbor nodes of the node, and the probability of each neighbor node being selected conforms to the calculated probability value; then, according to the obtained probability value and the walking parameter, the walking is started to obtain several walking paths; according to the walking path, the Word2vec method is called for training to obtain a word vector for the multi-label classification task of the knowledge network node;

[0034] The multi-element relationship reasoning component constructs possible candidate LE modes based on the combination of all edge types on the knowledge network; all instances corresponding to different candidate LE modes are found by traversing the whole graph, and the weight and value of different candidate LE modes are calculated to obtain a reliable LE mode by using a greedy algorithm; the reliable LE mode selected is used for matching in the knowledge network to obtain a reasoning result;

[0035] The dynamic knowledge reasoning component decouples the network topology in the form of time slices and accumulatively expands, sets position tags for each node and link edge in each time slice to indicate whether the network node and edge exist in the corresponding time slice, and discovers and reasons the knowledge group.

[0036] The information query method disclosed by the application is executed based on the information collaborative processing platform based on the knowledge graph as described above;

[0037] The information query method comprises the following steps:

[0038] Business data in the field of military intelligence is acquired from multiple data sources, and after cleaning and classification, the data is stored in the corresponding database; the business data in the field of military intelligence includes various military intelligence achievement data analyzed and researched in recent years and intelligence information reported by departments at all levels in real time; the multi-source heterogeneous business data in the field of military intelligence collected by the data collection module is uniformly represented; a semantic network knowledge base is generated according to the military intelligence field business data after format conversion, the semantic network knowledge base is dynamically expanded, and the repeated content and conflict content in the construction of the knowledge ontology are fused; the semantic network knowledge base includes four types of field knowledge ontology of person, organization, resource and military scene event, each field knowledge entity has corresponding attribute information, and each field knowledge entity includes an active state attribute;

[0039] Some words or phrases with greater relevance to the entity are acquired as basic keywords; based on the evolution trend of historical military strategy related information, the current period military strategy guidance information and the information query data of military experts, a core keyword set of the current period is constructed; then, according to the dependency relationship of syntactic analysis and the keyword co-occurrence frequency, the remaining basic keywords are layered and classified to construct a spatial distribution model between keywords and entities and between keywords and keywords;

[0040] The association relationship between the keywords and the knowledge ontology of each field is taken as a main feature, the feature vector of the knowledge ontology of each field is established, the similarity between the knowledge ontology of each field is calculated, and the relationship between the knowledge ontology of each field is obtained according to a general threshold value; starting from the keyword frequency and distance features, the hierarchical and sequential relationship between the knowledge entities of each field is mined, the semantic network of the evolution of the knowledge ontology is constructed, and the multi-source heterogeneous data in the semantic network knowledge base is subjected to semantic fusion;

[0041] Starting from the dynamic evolution of the knowledge ontology of each field and the mutual relationship, the upper and lower semantic reasoning of the fused semantic network knowledge base is carried out, the link between the two knowledge ontologies is found, the link extension is carried out, the potential semantics or more extensive relationship on the link is found, the evolution trend of the multi-element relationship between the person, organization, resource behavior sequence and military scene event is mined, and the multi-element knowledge semantic network is constructed; the decoupling of the network topology is accumulated and expanded in the form of time slices, the position label of each node and link edge in each time slice is set, the state transition of the network node and edge in the corresponding time slice is indicated, the knowledge group is found and reasoned, and the active state attribute of each field knowledge entity is updated;

[0042] The data classified by the data acquisition module is stored, and the message attribute of each database message is added;

[0043] According to the input entity keyword, the entity attribute information, entity graph information and entity relationship graph are extracted and displayed;

[0044] The related information of the event in the active state in the current period is managed, and the related information of all organizations is managed; according to the discovery and reasoning result of the knowledge group, the evolution state and evolution trend of the military strategy related information are discriminated from the aspects of person, organization, resource and military scene event, and the active state attribute of each field knowledge entity is updated, a plurality of key attention persons, key attention organizations, key attention resources and key attention events in the current period are screened out according to the updated active state attribute, and a military strategy graph is generated;

[0045] The evaluation result of the military strategy related information collected by the military experts is combined with the evolution state and evaluation result of the military strategy related information to dynamically update the military strategy guidance information of the next period.

[0046] Compared with the prior art, the beneficial effects of the present application are as follows:

[0047] The information collaborative processing platform based on the knowledge graph and the information query method of the application are developed from the actual military field business data, oriented by the real-time military strategy, research the military knowledge graph for the military intelligence information field, integrate a large number of scattered and isolated intelligence, realize the correlation of the military intelligence information from the semantic level, provide strong support for the comprehensive analysis, research and display of intelligence information, improve the analysis efficiency, realize the effective collaboration among the military strategy display, the dynamic adjustment of the military strategy and the real-time processing of intelligence data, and can provide the strong and complete military intelligence information query function. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 It is a structure schematic diagram of the information collaborative processing platform based on the knowledge graph of the application;

[0049] Figure 2 It is a principle diagram of entity recognition and attribute extraction;

[0050] Figure 3 It is a principle diagram of conflict resolution;

[0051] Figure 4 It is a structure schematic diagram of the spatial distribution model between the keywords and the entities and the keywords and the keywords;

[0052] Figure 5 It is a schematic diagram of the display result of the database module;

[0053] Figure 6 It is an entity extraction type schematic diagram;

[0054] Figure 7 It is an entity recognition result schematic diagram;

[0055] Figure 8 It is a multi-element relationship display result schematic diagram;

[0056] Figure 9 It is an event special topic library schematic diagram;

[0057] Figure 10 It is an organization library schematic diagram. DETAILED DESCRIPTION

[0058] The embodiments of the application are further described in detail below with reference to the accompanying drawings.

[0059] Reference Figure 1 The embodiment discloses an information collaborative processing platform based on a knowledge graph, the information collaborative processing platform comprises a data acquisition processing layer, a knowledge extraction intelligent analysis layer and a knowledge application decision layer;

[0060] The data acquisition processing layer comprises a data acquisition module, a data format conversion module and a knowledge base management module;

[0061] The data acquisition module obtains military intelligence field business data from multiple data sources, stores the classified data to the corresponding database, and the military intelligence field business data includes various types of military intelligence achievement data analyzed and researched in recent years and intelligence information reported by departments at all levels in real time; the data format conversion module is used for unified representation of the multi-source heterogeneous military intelligence field business data collected by the data acquisition module; the knowledge base management module is used for generating a semantic network knowledge base according to the military intelligence field business data after format conversion, dynamically expanding the semantic network knowledge base, and fusing repeated content and conflict content in the construction of the knowledge ontology; the semantic network knowledge base includes four types of field knowledge ontology of person, organization, resource and military scene event, each field knowledge entity has corresponding attribute information, and each field knowledge entity includes an active state attribute;

[0062] The knowledge extraction intelligent analysis layer includes a keyword extraction module, a keyword hierarchical module, a semantic fusion module and a knowledge ontology dynamic evolution module;

[0063] The keyword extraction module is used for extracting a plurality of words or phrases with greater relevance to entities as basic keywords;

[0064] The keyword hierarchical module constructs a core keyword set of the current period based on the evolution trend of historical military strategy related information, the current period military strategy guidance information and the information query data of military experts; then the remaining basic keywords are hierarchically and classified according to the dependency relationship of syntactic analysis and the keyword co-occurrence frequency, and the spatial distribution model between keywords and entities and keywords and keywords is constructed;

[0065] The semantic fusion module takes the association relationship between keywords and field knowledge ontology as the main feature, establishes the feature vector of the field knowledge ontology, calculates the similarity between the field knowledge ontologies, and obtains the relationship between the field knowledge ontologies according to the general threshold; starting from the keyword frequency and distance features, the hierarchical and sequential relationships between the field knowledge entities are mined, the knowledge ontology evolution semantic network is constructed, and the multi-source heterogeneous data in the semantic network knowledge base is semantically fused;

[0066] The knowledge ontology dynamic evolution module performs upper and lower semantic reasoning on the fused semantic network knowledge base from the dynamic evolution of each field knowledge ontology itself and mutual evolution, finds the link between two knowledge ontologies, extends the link, discovers potential semantics or more extensive relationships on the link, mines the evolution trend of the multi-element relationship between the person, organization, resource behavior sequence and military scene event, and constructs a multi-element knowledge semantic network; the network topology decoupling is accumulated and expanded in a cumulative manner by introducing a time slice, a position tag is set for each node and link edge in each time slice, the state transition of the network node and edge in the corresponding time slice is indicated, the knowledge group is discovered and reasoned, and the active state attribute of each field knowledge entity is updated;

[0067] The knowledge application decision layer includes a database module, an entity and relationship extraction module, an event topic library, an organization library, an evaluation result collection module, a military strategy guidance information update module and a military strategy graph generation module.

[0068] The database module is used for storing the data classified by the data collection module and adding a message attribute to each message in the database; the entity and relationship extraction module is used for extracting and displaying entity attribute information, entity graph information and entity relationship graph according to the input entity keyword; the event topic library is used for managing the related information of the event in the active state in the current period; the organization library is used for managing the related information of all organizations; the military strategy graph generation module is used for judging the evolution state and evolution trend of the military strategy related information from the aspects of person, organization, resource and military scene event according to the discovery and reasoning results of the knowledge group, updating the active state attribute of each field knowledge entity, screening a plurality of key attention persons, key attention organizations, key attention resources and key attention events in the current period according to the updated active state attribute, and generating a military strategy graph; the evaluation result collection module collects the evaluation results of military experts on the military strategy related information; and the military strategy guidance information update module dynamically updates the military strategy guidance information of the next period in combination with the evolution state and evaluation results of the military strategy related information.

[0069] I. Data collection and processing layer

[0070] The data collection processing layer is used for realizing collection, preprocessing and basic data analysis of multi-source heterogeneous data, the basic data sources mainly include mass military intelligence field business data accessed in real time and various intelligence achievements of analysis and research in recent years and intelligence information reported by troops at various levels, on the basis of which, the basic theory of the field knowledge ontology model is researched, various relations and relation definitions are constructed, and various data resources are organized, managed and stored around the field knowledge ontology model.

[0071] In order to effectively collect data, the application provides construction and display functions of a basic database, the basic database supports batch entry of inventory basic data, integration of basic data and trend intelligence data on the same platform, uploading, storage, classification management functions of basic data, display and viewing functions of basic data messages and full-text retrieval of basic database data and the like.

[0072] In order to solve the problem of inconsistent knowledge description in multi-source semantic network knowledge base fusion, a unified multi-source semantic knowledge base fusion framework is adopted. The fused semantic network knowledge base mainly consists of four parts: knowledge ontology / entity, category, attribute and attribute value. The entity as a leaf node of the entire knowledge base semantic network is an instance. The category hierarchical system consists of categories, sub(parent) categories and entities under the categories. The entity or category can belong to multiple categories or parent categories respectively. As shown in Table 1:

[0073] Table 1 Time sequence and space content model of knowledge ontology

[0074]

[0075] The entity is the smallest unit of various knowledge bases, is the naming of a thing and can belong to a category, and is a specific instance of the category.

[0076] The category also gives a simple document explanation in the semantic network knowledge base, and the parent class, child class and entity of the category are given.

[0077] The attribute is used to describe the entity and the category, and for each entity or category, the attribute is a term, category or field knowledge description obtained from the semantic network knowledge base.

[0078] The attribute value is a specific description of the attribute in the entity.

[0079] Based on the above four-tuples, this project presents the basic structure of a semantic knowledge base (SKB):

[0080] SKB = (C, E, A, V)

[0081] Where C = {c1, c2, ..., c} n} represents the category set, E = {e1, e2, ..., e m Let A = {a1, a2, ..., a} represent a set of entities. k} represents the set of entity attributes, where a i ∈A represents a basic attribute. This represents the value corresponding to the attribute. In the category set {C}, there are parent-child relationships between categories, i.e., c j ∈c s , indicating category c j It is category c s Subclasses of (Chile-Category).

[0082] Entity recognition classifies entities in text into a specific semantic type. This invention primarily employs three methods for entity recognition: dictionary-based, statistical, and rule-based methods. Dictionary-based methods mainly search for named entities in a lexicon through string matching. Rule-based methods incorporate lexical, syntactic, and semantic rules into the entity recognition process, then identify various types of named entities through rule matching. Statistical methods utilize manually labeled or raw corpora for training, first establishing a language model, then estimating model parameters on training data, which is beneficial for portability to different languages ​​and new domains. Statistical methods primarily utilize statistical models such as Hidden Markov Models, Maximum Entropy Models, Support Vector Machines, Conditional Random Fields (CRFs), and deep learning. Attribute extraction involves constructing an attribute table for each entity's semantic class and extracting attribute values. The main method for attribute extraction is pattern matching. See the diagrams for the principles of entity recognition and attribute extraction. Figure 2 In this invention, the semantic network knowledge base includes four domain knowledge ontology categories: people, organizations, resources, and military scenario events. Each domain category has subcategories, and in practical applications, further subdivision of these subcategories is required. These four domain knowledge ontology categories can be used to display four types of information in the final military strategy map. The domains are also closely related, and the selection of core keywords is also of positive significance.

[0083] The knowledge base management module includes an entity construction component, a semantic network knowledge base generation component, and a conflict fusion component;

[0084] The entity construction component is used to study the characteristics of the partially unstructured text, manually construct an initial military domain knowledge entity attribute library, and then use a deep learning LSTM sequence mining algorithm to establish word similarity to expand the military domain knowledge entity attribute library. Meanwhile, the unstructured text to be processed is annotated with entity and attribute by using a dictionary method combined with a part-of-speech attribute determination method.

[0085] In the aspect of entity and attribute feature extraction, the semantic network knowledge base generation component uses the keyword spatial position co-occurrence feature as a training set, trains the annotated entity and attribute data by using a deep learning model, selects an effective parameter model to extract new domain knowledge data, evaluates the effectiveness of the extracted entity and attribute, dynamically expands the evaluated effective entity and attribute to the military domain knowledge entity attribute library. On this basis, the corpus is continuously increased, the domain knowledge entity attribute library is continuously dynamically expanded, and the domain knowledge entity and attribute in a large amount of text are recognized by using a reinforcement machine learning method.

[0086] In the process of constructing the entity attribute set, due to the credibility and expression of each knowledge item corpus, conflicts may occur. The conflict fusion component analyzes the term conflict, semantic conflict and predicate conflict, and uses a hybrid method of logical tree fusion, frequency fusion and syntax fusion to select and fuse the repeated content and conflict content in the construction of the knowledge ontology. As shown in Figure 3 The conflict fusion component includes a term library, a predicate library, an ontology library, a term conflict processing unit, a semantic conflict processing unit, a predicate conflict processing unit and a fusion unit.

[0087] The term library and the predicate library respectively extract relevant knowledge ontology from the military field knowledge entity attribute library, respectively construct a term set, a predicate set and a semantic set and transmit to the ontology library; the ontology library analyzes the term set, the predicate set and the semantic set to obtain existing term conflicts, predicate conflicts and semantic conflicts; the term conflict processing unit calls a logical tree fusion model, a frequency fusion model and a syntax fusion model to process the term conflicts, the semantic conflict processing unit calls the syntax fusion model to process the semantic conflicts, and the predicate conflict processing unit calls the frequency fusion model and the syntax fusion model to process the predicates; the fusion unit integrates the processing results of the term conflict processing unit, the semantic conflict processing unit and the predicate conflict processing unit to generate a fused knowledge item. The logical tree fusion model fuses the conflicting items by judging the logical tree relationship between the two conflicting items; for example, the attribute attribute1 of the extracted entity Entity0 between the knowledge items is extracted to two values value1 and value2, without considering the time state migration between value1 and value2, the term logical conflict between value1 and value2, the project first judges the logical tree relationship between value1 and value2. If value1 contains value2, value1 is taken as the secondary attribute attribute1 of the attribute attribute1 of the entity Entity0; otherwise, the same processing is performed. The frequency fusion model adopts high-frequency co-occurrence knowledge items as the highest score output; the syntax fusion model uses a syntax dependency tree discriminator to judge the semantic space dependency relationship between entities, attribute words and attribute value words, and selects multiple items that produce conflicts to realize knowledge fusion, authentication and truth storage.

[0088] II. Knowledge extraction intelligent analysis layer

[0089] The knowledge extraction intelligent analysis layer provides a target-centered multi-mode data processing method and capability for the system, and provides an offline calculation, real-time calculation and graph calculation framework for various data analysis application requirements of the system in combination with the data storage model of the system; implements knowledge ontology and event extraction, network structure evolution, event correlation analysis and other processing services based on deep learning, constructs a field knowledge ontology library and a multi-field knowledge ontology network relationship; realizes behavior pattern discovery of the field knowledge ontology, constructs the relationship between multi-field knowledge ontologies, and realizes reasoning application in the dynamic evolution process of the field knowledge ontology. The knowledge extraction intelligent analysis layer includes a keyword extraction module, a keyword hierarchical module, a semantic fusion module and a knowledge ontology dynamic evolution module.

[0090] (2.1) Keyword extraction module

[0091] The keyword extraction module is mainly used for obtaining a plurality of words or phrases with relatively large relevance to the entity. First, a word set D-terms is obtained by segmenting each domain knowledge entity content document set D. The document D is mapped to a graph G, wherein the vertex is a word, and the edge represents the co-occurrence frequency of two words. Then, the BC (network node centrality) value between nodes is calculated, and the BC values of the words are sorted. Some words are merged into phrases by an n-gram algorithm, and finally a plurality of results with the highest ranking are output as keywords related to the domain knowledge entity.

[0092] (2.2) Keyword hierarchical module

[0093] In order to further show the hierarchical relationship of the extracted keywords, according to the regularity of the appearance of the keywords in the military domain knowledge entity document set, the keywords are divided into core keywords, important keywords and general keywords, and the relationship between the domain knowledge entity and the keywords is constructed. The spatial distribution model between the keywords and the entity and the keywords and the keywords obtained by construction is as shown in FIG. 2. Figure 4

[0094] In the present application, the guidance of the military strategy to the intelligence data aggregation is mainly realized by the selection of the core keywords. Specifically, the keyword hierarchical module selects a first keyword from the basic keywords based on the association relationship between the historical military strategy related information evolution trend and the keywords; obtains and analyzes the current period military strategy guidance information, and selects a second keyword from the basic keywords according to the military strategy guidance information; and comprehensively selects the first keyword and the second keyword to form a core keyword set of the current period. It should be noted that the number of the first keyword and the second keyword selected here is usually multiple.

[0095] ​The first keyword as a common keyword is greatly affected by the evolution trend and real-time evolution state of historical military strategy related information, is not affected by the subjective influence of military experts, and has higher objectivity. In order to ensure that the military staff can provide as complete intelligence information as possible, after calibrating the first keyword, a more extensive core keyword set can be obtained by training using an LSTM deep sequence neural network. The second keyword as a characteristic keyword has better relevance to the military strategy guidance information of the current period, so that the subsequent knowledge reasoning process can be more targeted to the real-time military strategy changes, and the coordination between military strategy and data processing is realized. As one of the preferred examples, the selection of the first keyword can be performed by setting keyword selection rules or introducing a machine learning algorithm. First, the evolution trend of historical military strategy related information is analyzed, important persons, organizations, resources or military scene events are screened, and then the first keyword is selected by combining the relevance between the important persons, organizations, resources or military scene events and the keyword. As for the selection of the second keyword, the viewing tendency data of military staff on military strategy related information, the evaluation results of military staff on military strategy related information, etc. can be collected, and by analyzing the keywords related to the tendency data or evaluation results, some second keywords can be selected or excluded by adjusting the priority. Thus, the core keywords obtained by combining the first keyword and the second keyword have objectivity and generality, and also have real-time and pertinence under the guidance of strategy, so that the subsequent knowledge reasoning process can also form a coordination with the change process of military strategy.

[0096] On the basis of the determination of the core keywords, the keywords extracted by the keyword extraction module are divided into important and general keywords related to the domain knowledge by the dependency relationship of the syntax analysis and the co-occurrence frequency of the keywords, and the spatial distribution model between the keywords and the entities and the keywords and the keywords is constructed.

[0097] (2.3) Semantic fusion module

[0098] The semantic fusion module includes a domain knowledge entity association relationship extraction component, a before-after causal relationship discrimination component and a semantic fusion component of multi-source heterogeneous data.

[0099] The domain knowledge entity association extraction component uses the association between keywords and domain knowledge entities as the main feature, establishes feature vectors for domain knowledge entities, calculates the pairwise similarity between domain knowledge entities, and obtains the relationships between domain knowledge entities based on a general threshold. Considering the sparsity of domain knowledge entities, a hierarchical training strategy, HAM (hierarchical abstract machines), is adopted to obtain models with different levels of accuracy. The top-level model is used to remove a large amount of irrelevant data, while the bottom-level models use a multi-classification method to accurately identify the associations between domain knowledge entities.

[0100] The prior causality discriminant component is used to determine the prior causal relationships in a domain knowledge entity network. Specifically, determining the prior relationships between domain knowledge entities that have already established connections can be treated as a binary classification problem, i.e., it is transformed into a feature selection and classification algorithm selection problem. The prior causality discriminant component mainly starts from keyword frequency and distance features. Considering that there may be correlations between the above feature vectors, for example, SVM can be used to identify the prior relationships between domain knowledge entities.

[0101] Given the definitions of entities and categories, the semantic fusion components of multi-source heterogeneous data, namely the entity similarity calculation model and the category network fusion model, fuse multiple semantic knowledge networks.

[0102] The entity similarity calculation model uses the following formula to fuse entities between the semantic network knowledge bases SKB1 and SKB2:

[0103] f:(C i E i A i V i )×(C j E j A j V j → (C, E, A, V)

[0104] Specifically, if the entity names in two knowledge bases are the same, then the two knowledge bases are considered to describe the same content; if the entity name in SKB1 is e i No matching entity name was found in SKB2. j However, it finds e that has the same domain knowledge description or a similarity greater than the first preset similarity threshold. j If the entity name in SKB1 is the same as the entity name in the other knowledge base, then the entities in the two knowledge bases are considered to be the same; if the entity name in SKB1 is the same as the entity name in the other knowledge base, then the entities in the two knowledge bases are considered to be the same. i In SKB2, only pairs of characters with a similarity less than the first preset similarity threshold can be found. jbut other attributes and attribute values are greater than the second preset similarity threshold, the entity in the two knowledge bases is considered to be the same; when the two entities are determined to be the same, e = e i or e = e j , C = (C i ∪C j ), A = (A i ∪A j ), wherein

[0105] When the two entities are determined to be the same or the two semantic network category nodes are fused, the category network fusion model is used to perform C = (C i ∪C j ); when the entity has only one parent category, whether the parent categories of the two knowledge bases are the same is determined:

[0106] If the category words of c i and c j are the same, the parent categories are determined to be the same;

[0107] If the category words of c i and c j are inconsistent, but the domain knowledge description content is the same or the similarity is greater than a third preset threshold, the categories in the two knowledge bases are considered to be the same; if the category words and the domain content description are inconsistent, the similarity of other attributes and attribute values of the category is greater than a fourth preset similarity threshold, and the categories in the two knowledge bases are considered to be the same; when the entity has multiple parent categories in different knowledge bases, and C i ∩C j is taken as the common parent category of the entity, and other parent categories call the depth inheritance algorithm and the fusion algorithm.

[0108] (2.4) Knowledge ontology dynamic evolution module

[0109] The reasoning technology of the semantic network knowledge can be divided into the reasoning of the upper and lower semantics of the unit knowledge node, the relationship reasoning of the multi-unit knowledge node and the knowledge change reasoning of the whole network dynamic evolution, from the knowledge ontology self-generation and the dynamic evolution between each other.

[0110] In the aspect of unit knowledge reasoning, the application improves the existing random walk algorithm, and constructs a semantic network reasoning algorithm for the field of military intelligence. It is assumed that the knowledge network data to be processed has N nodes, the current node of the walk process is C, and the next node in the walk path is selected from the neighbor nodes of C. It is assumed that C has E neighbor nodes, which are represented as:

[0111] neighbors(C) = {n1, n2, n3,..., n E}, 0≤E<N

[0112] At the same time, the neighbors(C) with the node C having a common label are expressed as:

[0113] common(C) = {m1, m2, m3,..., m k}, 0≤k≤E

[0114] Obviously, common(C) belongs to the subset of neighbors(C). Let D be selected as the next node of the node C in the walk, wherein D belongs to the set of neighbors(C). The probability that the node D belongs to the set of common(C) satisfies

[0115] P(D∈common(C)) = p, D∈neighbors(C)

[0116] wherein the probability p is a walk parameter set before the node walk starts. A new variable of the node C is calculated:

[0117]

[0118] A probability of being walked to is assigned to each neighbor node of the node C. The probability values are passed to the AliasMethod. The overall calculation process of the unit knowledge reasoning is as follows:

[0119] Step 1: Load the knowledge network data, and establish the data structure of the neighbor nodes and the label information corresponding to each node;

[0120] Step 2: For each node in the knowledge network, according to the label attributes of the node and its neighbor nodes, and the specified label information proportion adjustable parameter p, the probability values of the neighbor nodes of the node being walked to are calculated, and the AliasMethod is used to randomly select several times from the neighbor nodes of the node, and the probability of each neighbor node being selected conforms to the calculated probability value.

[0121] Step 3: According to the probability values obtained in the previous step and other walk parameters, such as the walk length, the number of times of starting the walk from each node, the walk is started, and a plurality of walk paths are obtained;

[0122] Step 4: According to the walk path, the Word2vec method is called for training, and the word vector is obtained;

[0123] Step 5: The classification task of the multi-label of the knowledge network node is performed, and the classification effect of the algorithm is verified.

[0124] The multi-element relationship reasoning can be specifically described as link extendable (LE) after finding the link between two knowledge ontologies, discovering the potential semantics or more extensive relationships on the link. The present application adopts the greedy algorithm in reinforcement learning to realize the fast convergence reasoning in multi-element relationship, and the specific steps are as follows:

[0125] 1) Construct the possible LE mode, for all edge types on the knowledge network GT / S, any combination of 3 types of edges can constitute a candidate LE model.

[0126] 2) Reinforcement learning realizes mode selection, finds all instances of different modes corresponding to the whole graph, calculates the weight (Weight) and value (Value) of different modes, and solves the reliable LE mode through the greedy algorithm.

[0127] 3) Match in the network using the selected mode to obtain the reasoning result.

[0128] The whole network dynamic evolution knowledge change reasoning, the knowledge discovery for military intelligence field must solve the high-speed change of knowledge, the present application adopts the Streaming stream processing framework of Spark platform to realize the dynamic reasoning and knowledge discovery for knowledge network. In order to reduce the incremental algorithm pressure of the platform to the sparse data stream, the time slice is introduced to decouple the network topology for cumulative expansion, through setting the position label of each node and link edge in each time slice, indicating whether the network node and edge exist or the state transition in the corresponding time slice, realizing the discovery and reasoning of knowledge group. The knowledge ontology dynamic evolution module also updates the active state attribute of each field knowledge entity, and the update rules of the active state of different field knowledge entities can be formulated in a self-defined manner, for example, for part of the weapons, as long as they are still in the research or existence state, they will always be in the active state, for part of the conference military scene events, after the conference ends, they can be adjusted to the non-active state. Preferably, different field knowledge entities have different active values, which are used to divide the key attention events and non-key attention events. For example, the evaluation results of military experts on military scene events can adjust the active value, the active value of part of resource entities is related to the attribute value, the active value of part of characters or organizations is related to the complexity of the associated relationship or the active value of the associated node, and so on.

[0129] The present application constructs a knowledge graph based on extracted entities, attributes, relationships, etc. For example, a semantic network based on a graph data structure composed of nodes (points) ("entities") and edges ("relationships") is constructed, a relationship network obtained by connecting all different types of information (persons, organizations, weapons, countries, etc.) together can be used to analyze and solve problems from the perspective of "relationships". In order to facilitate the operation of staff, the knowledge extraction intelligent analysis layer also provides some plug-ins, respectively used to support the storage, query and front-end display functions of the knowledge graph, support the automatic updating of the knowledge graph, and support the manual editing function of the knowledge graph.

[0130] III. Knowledge application decision layer

[0131] The knowledge application decision layer provides business support for knowledge application in the field of military intelligence. On the one hand, it provides military staff with business applications related to knowledge in the field of military intelligence, such as providing military strategy map display, intelligent search, ontology target portrait, relationship network analysis, time series analysis, knowledge recommendation, and other business service capabilities, providing historical event display function, providing data service and operation service support for upper layer application, etc. On the other hand, the knowledge application decision layer also needs to judge the evolution state and evolution trend of military strategy related information, collect real-time military strategy intention of military staff, dynamically update the next period of military strategy guidance information combined with the evolution state and evaluation results of military strategy related information, and automatically optimize the core keywords using the military strategy guidance information to guide the knowledge extraction intelligent analysis layer to optimize the knowledge reasoning process, so that the knowledge reasoning result is more in line with the real-time military strategy demand.

[0132] The knowledge application decision layer includes an evolution state discrimination module, a military strategy map generation module, an evaluation result collection module, and a military strategy guidance information updating module. The evolution state discrimination module discriminates the evolution state and evolution trend of military strategy related information from the aspects of persons, organizations, resources, and military scene events according to the discovery and reasoning results of the knowledge group. The military strategy map generation module generates a military strategy map according to the output results of the evolution state discrimination module. The evaluation result collection module collects the evaluation results of military experts on military strategy related information. The military strategy guidance information updating module dynamically updates the next period of military strategy guidance information combined with the evolution state and evaluation results of military strategy related information.

[0133] The knowledge application decision layer also includes a database module, an entity and relationship extraction module, an event topic library, an organization library, and other query modules.

[0134] For the application of military intelligence field knowledge, the application takes military strategy as the core of guidance, and researches from three aspects of knowledge query, knowledge recommendation and visualization display. Taking knowledge query as an example, the user's query of knowledge is often expressed in the form of a set of keywords, and the answer is often a set of connected nodes, which is represented as a subgraph matching the search content in the knowledge network. Due to the semantic completeness of keyword matching, a huge knowledge network has many subgraphs that meet the conditions, or has different topological structures, or has different keyword distribution. In the query search process, first, the content relevance of nodes and edges is calculated, a heuristic search algorithm is used to generate search results, and a sorting algorithm is used to sort the top-k subgraphs, and the final top-k search results are given. The knowledge application decision layer provides database, one-dimensional relationship query component, multi-dimensional relationship query component, event special topic library and organization library query modules according to the particularity of military strategy. Preferably, the entity and relationship extraction module further includes a one-dimensional relationship query component and a multi-dimensional relationship query component, which are used to provide one-dimensional relationship query and multi-dimensional relationship query functions.

[0135] The database module includes a database management component, a database query component and a database display component; the database management component is used to store the data classified by the data acquisition module into two types of databases, namely the basic database and the trend database, and add the storage time, the document date and the keyword attribute to each storage document; the database query component provides data query function, including document date time range plug-in, keyword query plug-in and database type, searches the document in the selected database through the user input document date range and keyword, and calls the database display component to push the search result to the display large screen. Figure 5 The display result of the database module is shown in the schematic diagram.

[0136] The entity and relationship extraction module includes an entity recognition component, an entity graph component and a relationship recognition component; the entity recognition component queries the related entity according to the input keyword, and pushes the related attribute of the query entity to the display large screen; the entity graph component includes a first entity name plug-in, and the entity graph component queries the related nodes of the entity in the knowledge graph according to the input single entity name, obtains and displays the associated attribute and relationship query result; the relationship recognition component includes a relationship selection plug-in and a plurality of second entity name plug-ins, and the relationship recognition component inquires and displays the knowledge graph of the specified relationship between the input multiple entities through multi-layer penetration between multiple entities according to the input multi-dimensional relationship type; the specified relationship includes the association relationship, the associated organization and the associated event between entities.

[0137] For example, input a single entity to a unary relation query component, and the related nodes of the entity in the knowledge graph can be queried to obtain query results such as associated attributes and relations. For example, unary relation (person-person relation) query, unary relation (organization-person relation) query, unary relation (person-organization relation) query. Input multiple entities to a multi-relation query component, and after multi-layer penetration between the multiple entities, the associated relations, associated organizations, associated events and the like between the entities can be obtained. For example, multi-relation (person-person relation) query, multi-relation (person-organization relation) query, and the like.

[0138] The following will be described in combination with Figure 6 and Figure 7 The extraction results of entities, attributes and the like are exemplified.

[0139] Entity extraction types: person, organization, military unit, country, city, region, weapon, and the like.

[0140] Attribute extraction types:

[0141] Person attribute extraction types: Chinese name, English name, nationality, date of birth, place of birth, gender, position, education, religious belief, and the like; organization attribute extraction types: nature, establishment time, and establishment place.

[0142] Relation extraction types:

[0143] Person-to-person relations:

[0144] < Person > colleague < person >, < person > friend < person >, < person > classmate < person >, < person > fellow villager < person >, < person > superior < person >, < person > partner < person >, < person > cooperation < person >, < person > support < person >, < person > sponsor < person >.

[0145] Person-to-organization relations:

[0146] < Person > serves < organization >, < person > head < organization >, < person > core member < organization >, < person > member < organization >, < person > member < organization >, < person > support < organization >, < person > sponsor < organization >, < person > create < organization >, < person > cooperation < organization >.

[0147] Organization-to-organization relations:

[0148] < Organization > cooperation < organization >, < organization > sponsor < organization >, < organization > subordinate < organization >, < organization > competitive relation < organization >, < organization > hostile relation < organization >, < organization > support < organization >.

[0149] Person-to-country relations:

[0150] <PERSON> was born in <COUNTRY>, <PERSON> is a citizen of <COUNTRY>, <PERSON> resides in <COUNTRY>, <PERSON> studied in <COUNTRY>, <PERSON> works in <COUNTRY>, <PERSON> plays for <COUNTRY>, <PERSON> visited <COUNTRY>.

[0151] Relationships of a person to a military unit:

[0152] <PERSON> is a member of <MILITARY UNIT>, <PERSON> plays for <MILITARY UNIT>, <PERSON> visited <MILITARY UNIT>.

[0153] Relationships of a person to a city:

[0154] <PERSON> was born in <CITY>, <PERSON> resides in <CITY>, <PERSON> works in <CITY>, <PERSON> visited <CITY>.

[0155] Relationships of a weapon to a country:

[0156] <COUNTRY> developed <WEAPON>, <COUNTRY> possesses <WEAPON>, <COUNTRY> purchased <WEAPON>, <COUNTRY> captured <WEAPON>, <COUNTRY> intercepted <WEAPON>, <COUNTRY> used <WEAPON>, <COUNTRY> destroyed <WEAPON>.

[0157] Existing multi-relational query is often to calculate the maximum weight subgraph of the keyword set input by the user in the entire field knowledge network. Taking two keywords as an example, it is to calculate the shortest path or the optimal sense path between two nodes. In the graph, there are two cases: one is that although a node is connected with many nodes, the node and its closest node have extremely small relevance; the other is that although the distance between two nodes is long, the node and the node content on the path have extremely high similarity, and the comprehensive calculation of these nodes will have high relevance. Therefore, after calculating the shortest path between two nodes, the present application adopts the Steiner node expression method to judge the potential hub node (HITS algorithm can be used to calculate the hub value of each node) on the shortest path. If there is a hub node, the second shortest path is searched, and if the next shortest path does not exist, the last shortest path containing the hub node is used. This method can relatively make up for the deficiency of directly calculating the shortest path between two nodes. Figure 8 To show the result of multi-relational query.

[0158] See Figure 9The event topic database analyzes key events from multiple dimensions, including basic event information, event introduction, development history, related messages, relationship graphs, and trending keywords. It supports the addition and configuration management of key events, summarizing and analyzing different event topics arising from the evolution of events, aggregating related messages and performing timeline analysis based on event topics, and viewing related messages for event topics. Specifically, the event topic database includes an event analysis component, an event addition component, an event association component, a related message aggregation component, and a related message query component. The event analysis component analyzes key events from multiple dimensions, including basic event information, event introduction, development history, related messages, relationship graphs, and trending keywords. The event addition component allows manual addition of key events. The event association component summarizes and analyzes different event topics arising from the evolution of events. The related message aggregation component aggregates related messages for event topics according to a timeline. The related message query component provides the function of viewing related messages.

[0159] See Figure 10 The organization library manages key organizations, such as terrorist organizations, military organizations, and international organizations, from dimensions such as basic organizational information, organizational introduction, organizational history, organizational structure, and relationship graph. It also supports the addition and configuration management of organizations of interest. Specifically, the organization library includes components for basic organizational information management, organizational history management, related event management, related message management, and relationship graph generation. The basic organizational information management component manages the basic information of all organizational entities in the library; the organizational history management component manages the changes in the organization's personnel structure and resource information; the related event management component adds event types related to all organizational entities in real time based on knowledge reasoning results; the related message management component manages all messages related to the organization; and the relationship graph generation component organizes and generates a binary relationship knowledge graph related to the organization. Figure 5 , Figures 6 to 10 The text in this document is for illustrative purposes only and is not necessarily related to the technical principles of this application.

[0160] Those skilled in the art will appreciate that embodiments of the application can be readily used as software, hardware, or a combination of software and hardware. In one embodiment, the application can be implemented in software and / or firmware. In addition, those skilled in the art will further appreciate that the application can be implemented as a method, apparatus, or computer program product. Therefore, embodiments of the application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects all generally referred to herein as a "circuit" or "module." Furthermore, embodiments of the application can take the form of a computer program product on a computer-readable storage medium having computer program code embodied in the storage medium. The computer program code can cause a computer, processor, or other programmable data processing apparatus to effect the steps in the embodiments of the application as set forth in the description below.

[0161] The embodiments of methods, apparatuses (systems) and computer program products of the application are described herein with reference to flowchart and / or block diagrams illustrations of the methods, apparatuses (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart and / or block diagrams illustrations, and combinations of blocks in the flowchart and / or block diagrams illustrations, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing element or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block(s) or combinations thereof. Figure 1 one or more functions specified in the flowchart and / or block diagram block(s) or combinations thereof. Figure 1 one or more functions specified in the flowchart and / or block diagram block(s) or combinations thereof.

[0162] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flowchart and / or block diagram block(s) or combinations thereof. Figure 1 one or more functions specified in the flowchart and / or block diagram block(s) or combinations thereof. Figure 1 one or more functions specified in the flowchart and / or block diagram block(s) or combinations thereof.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flowchart and / or block diagram block(s) or combinations thereof. Figure 1 one or more functions specified in the flowchart and / or block diagram block(s) or combinations thereof. Figure 1 one or more functions specified in the flowchart and / or block diagram block(s) or combinations thereof.

[0164] While the preferred embodiments of the application have been described, additional variations and modifications can be employed by those skilled in the art. Therefore, the appended claims are intended to cover all such variations and modifications as falling within the scope of the application.

[0165] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A knowledge graph-based information collaborative processing platform, characterized in that, The information collaborative processing platform comprises a data acquisition processing layer, a knowledge extraction intelligent analysis layer and a knowledge application decision layer; The data acquisition processing layer comprises a data acquisition module, a data format conversion module and a knowledge base management module; The data acquisition module acquires military intelligence field business data from multiple data sources, stores the classified data into corresponding databases, and the military intelligence field business data comprises various military intelligence achievement data of recent years and real-time accessed intelligence information reported by departments at all levels; the data format conversion module is used for uniformly representing the multi-source heterogeneous military intelligence field business data collected by the data acquisition module; the knowledge base management module is used for generating a semantic network knowledge base according to the format-converted military intelligence field business data, dynamically expanding the semantic network knowledge base, and fusing repeated contents and conflict contents in the construction of the knowledge ontology; the semantic network knowledge base comprises four types of field knowledge ontology, i.e., person, organization, resource and military scene event, each field knowledge entity has corresponding attribute information, and each field knowledge entity comprises an active state attribute; The knowledge extraction intelligent analysis layer comprises a keyword extraction module, a keyword hierarchical module, a semantic fusion module and a knowledge ontology dynamic evolution module; The keyword extraction module is used for taking several words or phrases with relatively large entity correlation as basic keywords; The keyword hierarchical module constructs a core keyword set of the current period based on the evolution trend of historical military strategy related information, current period military strategy guidance information and information query data of military experts, and then classifies and hierarchically classifies the remaining basic keywords according to the dependency relationship of syntactic analysis and keyword co-occurrence frequency, to construct a spatial distribution model between keywords and entities and between keywords and keywords; The semantic fusion module takes the association relationship between keywords and field knowledge ontology as a feature, establishes a feature vector of the field knowledge ontology, calculates the similarity between two field knowledge ontologies, obtains the relationship between the field knowledge ontologies according to a general threshold, mines the hierarchy and sequence relationship between field knowledge entities from keyword frequency and distance features, constructs a knowledge ontology evolution semantic network, and performs semantic fusion on multi-source heterogeneous data in the semantic network knowledge base; The knowledge ontology dynamic evolution module performs upper and lower semantic reasoning on the fused semantic network knowledge base from the dynamic evolution of the field knowledge ontology itself and mutual evolution, finds the link between two knowledge ontologies after link extension, discovers potential semantics or more extensive relationships on the link, mines the evolution trend of the multi-element relationship between the person, organization, resource behavior sequence and military scene event, constructs a multi-element knowledge semantic network, introduces a time slice to decouple the network topology and cumulatively expand, sets a position label for each node and link edge in each time slice, indicates whether the network node and edge exist in the state transition in the corresponding time slice, discovers and reasons the knowledge group, and updates the active state attribute of each field knowledge entity. The knowledge application decision layer comprises a database module, an entity and relationship extraction module, an event topic database, an organization database, an evaluation result collection module, a military strategy guidance information updating module and a military strategy graph generation module; The database module is used for storing the data classified by the data collection module and adding message attributes to each message in the database; the entity and relationship extraction module is used for extracting and displaying entity attribute information, entity graph information and entity relationship graph according to the input entity keywords; the event topic database is used for managing the information related to the events in the active state in the current period; The organization database is used for managing the information related to all organizations; the military strategy graph generation module is used for determining the evolution state and evolution trend of the military strategy related information from the aspects of person, organization, resource and military scene event according to the discovery and reasoning results of the knowledge group, updating the active state attribute of each field knowledge entity, screening a plurality of key attention persons, key attention organizations, key attention resources and key attention events in the current period according to the updated active state attribute, and generating a military strategy graph; the evaluation result collection module collects the evaluation results of the military experts on the military strategy related information; and the military strategy guidance information updating module dynamically updates the military strategy guidance information in the next period in combination with the evolution state of the military strategy related information and the evaluation results. 2.The knowledge graph based information collaborative processing platform according to claim 1, characterized in that, The database module comprises a database management component, a database query component and a database display component; the database management component is used for storing the data classified by the data collection module into two types of databases, i.e., a basic database and a trend database, adding the storage time, message date and keyword attribute to each message in the database; the database query component provides a data query function, including a message date time range plug-in, a keyword query plug-in and a database type, searches the messages in the selected database through the input message date range and keyword, and calls the database display component to push the search results to the display large screen. 3.The knowledge graph based information collaborative processing platform according to claim 1, characterized in that, The entity and relationship extraction module comprises an entity recognition component, an entity graph component and a relationship recognition component; the entity recognition component queries the related entities according to the input keywords and pushes the related attributes of the queried entities to the display large screen; the entity graph component comprises a first entity name plug-in, and the entity graph component queries the related nodes of the entity in the knowledge graph according to the input single entity name, obtains and displays the associated attributes and relationship query results; The relationship recognition component comprises a relationship selection plug-in and a plurality of second entity name plug-ins, the relationship recognition component inquires the specified relationship knowledge graph between the input multiple entities through multi-layer penetration between multiple entities according to the input multiple relationship types, and displays the specified relationship knowledge graph; the specified relationship comprises the association relationship, associated organization and associated event between entities. 4.The knowledge graph based information collaborative processing platform according to claim 1, characterized in that, The event topic library comprises an event analysis component, an event adding component, an event correlation component, a correlation message aggregation component and a correlation message query component; the event analysis component analyzes each key focus event from multiple dimensions including event basic information, event introduction, development history, correlation message, relationship graph and event hot words; the event adding component is used for manually adding key focus events; the event correlation component is used for summarizing and analyzing different event topics generated by event development and evolution; the correlation message aggregation component is used for aggregating event topic related messages according to a time axis; and the correlation message query component is used for providing a correlation message viewing function. 5.The knowledge graph based information collaborative processing platform according to claim 1, characterized in that, The organization library comprises an organization basic information management component, an organization experience management component, a correlation event management component, a correlation message management component and a relationship graph generation component; the organization basic information management component is used for managing the basic information of all organization entities in the organization library; the organization experience management component is used for managing the relevant person structure and resource information change process of the organization; the correlation event management component is used for adding event types related to all organization entities in real time according to knowledge reasoning results; the correlation message management component is used for managing all messages related to the organization; and the relationship graph generation component is used for sorting and generating a binary relationship knowledge graph related to the organization. 6.The knowledge graph based information collaborative processing platform according to claim 1, characterized in that, The knowledge base management module comprises an entity construction component, a semantic network knowledge base generation component and a conflict fusion component; The entity construction component is used for researching the characteristics of part of unstructured texts, manually constructing an initial military field knowledge entity attribute library, expanding the military field knowledge entity attribute library based on word similarity, and labeling entities and attributes in unstructured texts by using a dictionary method combined with a part of speech attribute determination method; The semantic network knowledge base generation component takes the spatial position co-occurrence characteristics of keywords as a training set, trains the labeled entity and attribute data by using a deep learning model, extracts new field knowledge data by using the trained deep learning model, evaluates the effectiveness of the extracted entities and attributes, dynamically expands the evaluated effective entities and attributes to the military field knowledge entity attribute library, and identifies field knowledge entities and attributes in massive texts based on a reinforcement machine learning method; The conflict fusion component analyzes term conflicts, semantic conflicts and predicate conflicts, and adopts a hybrid method of logic tree fusion, frequency fusion and syntax fusion to select and fuse repeated contents and conflict contents in the construction of the knowledge ontology. 7.The knowledge graph-based information collaborative processing platform according to claim 6, characterized in that, The conflict fusion component comprises a term library, a predicate library, an ontology library, a term conflict processing unit, a semantic conflict processing unit, a predicate conflict processing unit and a fusion unit. The term library and the predicate library respectively extract relevant knowledge ontology from the military field knowledge entity attribute library, respectively construct a term set, a predicate set and a semantic set and transmit to the ontology library; the ontology library analyzes the term set, the predicate set and the semantic set, and obtains existing term conflicts, predicate conflicts and semantic conflicts; the term conflict processing unit calls a logical tree fusion model, a frequency fusion model and a syntax fusion model to process the term conflicts, the semantic conflict processing unit calls the syntax fusion model to process the semantic conflicts, and the predicate conflict processing unit calls the frequency fusion model and the syntax fusion model to process the predicates; The fusion unit integrates the processing results of the term conflict processing unit, the semantic conflict processing unit and the predicate conflict processing unit to generate a fused knowledge item; The logical tree fusion model fuses the conflicting items by judging the logical tree relationship of the two conflicting items; The frequency fusion model uses high-frequency co-occurring knowledge items as the highest score output; the syntax fusion model uses a syntax dependency tree discriminator to judge the semantic space dependency relationship between entities, attribute words and attribute value words, and selects multiple items that produce conflicts. 8.The knowledge graph based information collaborative processing platform according to claim 1, characterized in that, The keyword hierarchical module filters the entities in an active state in the current period based on the evolution trend of historical military strategy related information, and selects a first keyword related to the entity from the basic keywords according to the association relationship between each entity and the keyword; obtains and analyzes the military strategy guidance information of the current period, selects a second keyword related to the entity from the basic keywords according to the entity corresponding to the military strategy guidance information; and integrates the first keyword and the second keyword to form a core keyword set of the current period. 9.The knowledge graph based information collaborative processing platform according to claim 1, characterized in that, The knowledge ontology dynamic evolution module includes a unit knowledge reasoning component, a multi-element relationship reasoning component and a dynamic knowledge reasoning component; The unit knowledge reasoning component loads knowledge network data, establishes a data structure of neighbor nodes and label information corresponding to each node, calculates the probability value of the neighbor nodes of each node in the knowledge network being walked to according to the label attributes of the node and its neighbor nodes and a specified label information proportion adjustable parameter p, and randomly selects a number of times from the neighbor nodes of the node, and the probability of each neighbor node being selected conforms to the calculated probability value; then, according to the obtained probability value and a walking parameter, the walking is started to obtain a plurality of walking paths; According to the walking path, a Word2vec method is called for training to obtain a word vector for a multi-label classification task of the knowledge network node; The multi-element relationship reasoning component constructs possible candidate LE modes based on the combination of all edge types on the knowledge network, finds all instances corresponding to different candidate LE modes by traversing the whole graph, calculates the weight and value of different candidate LE modes, and solves the credible LE mode through a greedy algorithm; the selected credible LE mode is matched in the knowledge network to obtain a reasoning result. The dynamic knowledge reasoning component introduces time slices in a manner that accumulatively expands the decoupling of network topology, sets position tags for each node and link edge in each time slice, indicates whether there is a state transition of the network node and edge in the corresponding time slice, discovers and reasons the knowledge group. 10.A knowledge graph based information query method, characterized in that, The information query method is executed based on the information collaborative processing platform based on the knowledge graph as claimed in any one of claims 1-9; The information query method comprises the following steps: Business data in the field of military intelligence is acquired from multiple data sources, stored in corresponding databases after cleaning and classification, and includes various types of military intelligence achievement data analyzed and researched in recent years and intelligence information reported by departments at all levels in real time; the multiple-source heterogeneous business data in the field of military intelligence collected by the data collection module is uniformly represented; a semantic network knowledge base is generated according to the military intelligence field business data after format conversion, and the semantic network knowledge base is dynamically expanded, while the repeated content and conflicting content in the construction of the knowledge ontology are fused; the semantic network knowledge base includes four types of field knowledge ontology of person, organization, resource and military scene event, each field knowledge entity has corresponding attribute information, and each field knowledge entity includes an active state attribute; Some words or phrases with relatively large relevance to the entity are taken as basic keywords; a core keyword set of the current period is constructed based on the evolution trend of historical military strategy related information, military strategy guidance information of the current period and information query data of military experts; the remaining basic keywords are layered and classified according to the dependency relationship of syntactic analysis and the co-occurrence frequency of keywords, and a spatial distribution model between keywords and entities and between keywords and keywords is constructed; The association relationship between the keywords and the field knowledge ontology is taken as a feature, a feature vector of the field knowledge ontology is established, the similarity between the field knowledge ontologies is calculated, and the relationship between the field knowledge ontologies is obtained according to a general threshold value; starting from the keyword frequency and distance features, the hierarchy and sequence relationship between the field knowledge entities are mined, an ontology evolution semantic network is constructed, and the multiple-source heterogeneous data in the semantic network knowledge base is semantically fused; Starting from the dynamic evolution of the field knowledge ontology itself and each other, the upper and lower semantic reasoning of the fused semantic network knowledge base is performed, the link between the two knowledge ontologies is found, the link is extended, the potential semantics or more extensive relationship on the link is found, the evolution trend of the multi-element relationship between the person, organization, resource behavior sequence and military scene event is mined, and a multi-element knowledge semantic network is constructed; the decoupling of network topology is accumulatively expanded in a manner of introducing time slices, position tags are set for each node and link edge in each time slice, it is indicated whether there is a state transition of the network node and edge in the corresponding time slice, the knowledge group is discovered and reasoned, and the active state attribute of each field knowledge entity is updated. The data collection module stores the classified data and adds message attributes to each message in the database; manages the information related to events in the current period that are in an active state; manages the information related to all organizations; According to the input entity keyword, the entity attribute information, entity graph information and entity relationship graph are extracted and displayed; The information related to events in the current period that are in an active state is managed; the information related to all organizations is managed; according to the discovery and reasoning results of the knowledge group, the evolution state and evolution trend of the military strategy related information are judged from the aspects of characters, organizations, resources and military scene events, and the active state attributes of each field knowledge entity are updated; according to the updated active state attributes, a number of key attention characters, key attention organizations, key attention resources and key attention events in the current period are screened out, and a military strategy graph is generated; The evaluation results of military experts on the military strategy related information are collected, and the evolution state and evaluation results of the military strategy related information are combined to dynamically update the military strategy guidance information in the next period.

Images