CN114896387A - Military intelligence analysis visualization method and device and computer readable storage medium - Google Patents
Military intelligence analysis visualization method and device and computer readable storage medium Download PDFInfo
- Publication number
- CN114896387A CN114896387A CN202210054358.7A CN202210054358A CN114896387A CN 114896387 A CN114896387 A CN 114896387A CN 202210054358 A CN202210054358 A CN 202210054358A CN 114896387 A CN114896387 A CN 114896387A
- Authority
- CN
- China
- Prior art keywords
- event
- text
- node
- class
- military
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/34—Browsing; Visualisation therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Animal Behavior & Ethology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Machine Translation (AREA)
Abstract
The invention relates to a military intelligence analysis visualization method, which comprises the steps of obtaining and preprocessing event texts in various fields, dividing event classes of the event texts according to an event ontology base, extracting event elements of the event texts, extracting main contents and event class relations of the event texts according to the event elements, conducting prediction and derivation on event evolution paths through a Bayesian network classification model, and visually displaying the event class relations and the event evolution paths. Therefore, the intelligence event can be shown in the form of a graph as much as possible, the relevant influence brought by the event is predicted through a machine learning classification model, and customized domain event content is pushed for a corresponding user.
Description
Technical Field
The invention relates to the field of information analysis, in particular to a military information analysis visualization method and device and a computer readable storage medium.
Background
With the diversification of information, people are more and more convenient to obtain data, but just because of the diversification of the data, people cannot obtain data in a field concerned by people in a shorter time, and people hope to integrate the existing data information in the field concerned, can inquire the information of the correlation of the existing information, and mine the causal relationship in the data, sort the development venation of events and analyze the development rule of the events by combining the current data with the historical data. And it is desirable to be able to dynamically predict the trend of the evolution of future events. Therefore, the overall relationship network of the events and the possible evolution trend can be completely understood, and the data of the intelligence can provide important reference values for related intelligence personnel.
In the existing research, only a case map is simply constructed on a specific occurred event in a certain field, and the data is structured in detail and the data association is established on the structured data under an appointed framework. And then the data is visually displayed. Thereby building a summary of the particular event. For example, a case map of a certain internet public sentiment event is constructed, and now, intelligence analysis is only performed on intelligence of a specific field and possibly a specific event type, and a user cannot customize a field event concerned by the user, and analyze evolution of the concerned event and a related influence range.
Disclosure of Invention
In order to solve the above problems in the prior art, the invention provides a military intelligence analysis visualization method, which comprises the following steps:
s1, obtaining an event text of a selected military field, preprocessing the event text, dividing the event class of the event text according to a pre-established event ontology base and extracting event elements of the event text;
s2, extracting the main content and the event type relation of the event text according to the event elements to obtain an event training set;
s3, inputting the event training set into a pre-trained Bayesian network classification model to obtain an event distribution probability, and predicting and deducing an event evolution path according to the event distribution probability;
and S4, visually displaying the event class relationship and the event evolution path.
The invention can be improved as follows on the basis of the scheme.
Further, the event ontology library is pre-established by the following steps:
s101, acquiring a corresponding event text, performing part-of-speech tagging on content in the event text, and constructing a corresponding corpus;
s102, extracting event elements, event types and event type relations in an event text according to the corresponding corpus, and constructing a corresponding event ontology model of the military field;
s103, summarizing the event ontology models in the military affair field to obtain the event ontology library, and constructing a corresponding near-sense word library and a trigger word library according to the event class.
Further, the event elements in S102 are ontologies constituting an event text, and the event elements in S102 include a time element T, an environment element V, an object element O, a process state element P, a language expression element L, and an action element a; the event class is used for representing an event type; the event class relationship is used for representing the event relationship among the events.
Further, the obtaining of the event text in S1, and the preprocessing the event text includes: and performing word segmentation processing and format normalization processing on the event text, and giving corresponding part-of-speech marks to all words in the text according to grammar rules of the language to which the event text belongs.
Further, the S1 specifically includes: and dividing the event class of the event text according to the event ontology library and/or the trigger thesaurus and the synonym library, and extracting corresponding event elements according to an event element template corresponding to the event class of the event text.
Further, the step S2 is specifically to perform semantic analysis on the event text and extract semantic relationships between event elements and other event elements, analyze and label the event text according to the type of the customized event type relationship and extract corresponding event type relationships, where the semantic relationships are used to perform calculation and inference based on the event element information to obtain the main content of the event text.
Further, the S4 specifically includes the following steps: s401, establishing nodes according to the event ontology model;
s402, performing supplementary explanation on the attribute of the node according to the event element, labeling the event class corresponding to the node through a node label, connecting each node according to the event class relation of the node, and completing the establishment of a graph database;
and S403, analyzing the correlation among the nodes through a Bayesian network classification model to obtain the probability distribution of the node direction, labeling the node path according to the maximum probability node direction, and visually displaying the nodes, the node attributes, the node labels and the node paths.
Further, all the nodes are visually displayed according to event texts and/or trigger words input by users, the relevant nodes of the similar meaning word retrieval and the nodes with event type relations.
Further, the event type relation comprises a geographic space relation, all the nodes are visually displayed based on the geographic space, and corresponding data operation portals are provided.
Further, the user interest is judged by comparing the similarity of the event class and the retrieval result in the user retrieval history according to a personalized search algorithm, and then the search results are sequenced.
It is another object of the present invention to provide a computer readable storage medium having at least one program stored thereon, the at least one program being executed by the processor to implement the military intelligence analysis visualization method described above.
It is still another object of the present invention to provide a military intelligence analysis visualization apparatus, which comprises a memory and a processor, wherein the memory stores at least one program, and the at least one program is executed by the processor to realize the military intelligence analysis visualization method.
The invention has the advantages that event ontology models in different fields are built, unstructured information collected on a network is processed by a natural language processing technology (NLP), generated events are subjected to characteristic word extraction, text main content generation, event type relation recognition, different attribute recognition corresponding to different types of events and corresponding ontology model adaptation, entities of semantic units built from layers with different granularities and relevant relevance relation analysis are used for displaying the information events in a graph mode as much as possible, relevant influences brought by the events are predicted by a machine learning classification model, and customized field event contents are pushed for corresponding users.
Drawings
FIG. 1 is a flow chart of a military intelligence analysis visualization method of an embodiment of the present invention;
FIG. 2 is a flowchart of a corpus construction method according to an embodiment of the present invention;
FIG. 3 is a flowchart of a method for visualization via a graph database, in accordance with an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a feedback optimization mechanism according to an embodiment of the present invention.
Detailed Description
The principles and features of this invention are described below in conjunction with the following drawings, which are set forth by way of illustration only and are not intended to limit the scope of the invention.
As shown in fig. 1, the present invention provides a method for analyzing and visualizing military intelligence, comprising the following steps,
s1, obtaining the event text of the selected military field, preprocessing the event text, dividing the event class of the event text according to the pre-established event ontology base and extracting the event elements of the event text;
s2, extracting the main content and the event type relation of the event text according to the event elements to obtain an event training set;
s3, inputting the event training set into a pre-trained Bayesian network classification model to obtain an event distribution probability, and predicting and deducing an event evolution path according to the event distribution probability;
and S4, visually displaying the event class relationship and the event evolution path.
The invention constructs event ontology models in different fields, processes collected unstructured information through a natural language processing technology (NLP), extracts characteristic words, generates main text contents and identifies event relations of event texts, identifies event classes to which different types of events belong, performs adaptation extraction processing on corresponding event ontology models, constructs entities of semantic units from layers with different granularities and analyzes events with associated event relations, so that the events of the events can be displayed in a graph form as much as possible, and predicts related influences brought by the events, development directions and the like through a machine learning classification model.
In this embodiment, as shown in fig. 2, in step S101, event texts in the military field in the selected military field are obtained, an event ontology library of the event ontology is built in different fields, and by obtaining relevant event texts in the military field, content in the event texts is labeled, so as to build a corresponding corpus, such as a cec (chinese emery corpus) corpus. In step S102, according to the corpus corresponding to each field, the event elements, the event classes, and the event class relationships in the event text are extracted, and an event ontology model of each field is constructed. Extracting an event and event elements from an event text, wherein the event elements comprise a time element T, an environment element V, an object element O, a process state element P, a language expression element L and an action element A, the event can be represented as e, and the event elements can be defined as a six-tuple of (A, O, T, V, P, L); where a (action element) is typically used to represent the course of a change in an event and its characteristics, is a description of the degree, manner, method, tool, etc., such as how fast or slow a movement is, what is used or according to. And O (object element) refers to an engagement object of an event, including all roles engaged in the event, and the objects may be an actor (subject) and an actor (object) of an action, respectively. T (time element) is the time period of the event occurrence, from the start point of the event occurrence to the end point of the event end. And V (environmental element) represents the place where the event occurs and its environmental characteristics, etc. For example, swimming in the sea, the environment is the sea, and the site environment is characterized by water. P (process state element) is then composed of a precondition for the occurrence of an event, an intermediate assertion, and a postcondition. Preconditions refer to constraints that the elements should or may satisfy in order to perform the event, and they may be triggers for the occurrence of the event; the intermediate assertion refers to the condition that each element of the intermediate state in the event generation process meets; after an event occurs, each element of the event causes a change or a transition in the state of each element, and the result of the change or transition becomes a post-condition of the event. And finally, L (language expression element) is a language expression rule of the event, and comprises a core word set, core word expression, core word collocation and the like. The core words are the common notational vocabulary of events in sentences. The core word expression is the position relation between the expression of each element in the sentence and the core word. The core word collocation refers to the inherent collocation of the core word and other words. The representation of the event in different language categories, such as chinese, english, french, etc., may be based on the event.
The event ontology is used for researching the composition of events and the relationship among the events, an event ontology model is constructed to form an event ontology library after the events are induced and summarized, event classification relationship extraction is carried out on related texts in different fields, the event class, event element and event class relationship obtained through analysis are induced, a formal description language is obtained through abstraction and formalization, the existing formal language such as description logic, Z language and the like is comprehensively analyzed, and finally the description logic is used as a formal description basis of the event ontology model, or the event related element described based on semantic texts is converted into keyword-based description through defining keywords of the event ontology; the event classes, event elements and event relations which exist in the corpus are formally described, and the formally described events are stored in the event ontology in the format of an XML document.
Because the same event is often described in one or more related texts in any field, because the expression means and methods in each text are inconsistent, a plurality of events which are uniform and have similar expression modes are merged and fused from different texts, the extracted specific events can be referred to and resolved into the same abstract event, when a plurality of event expressions point to the same natural event, the event table complaints are considered to have the same reference relationship, the similarity of the contexts of the plurality of related events is tested according to a model, the similarity of the events is calculated, and the event classes are divided according to the common attribute and the similarity of the events.
In this embodiment, as shown in fig. 2, in step S103, the event ontology model in the military field is summarized to obtain the event ontology library, and a corresponding word thesaurus and a corresponding word library of trigger words are constructed according to the event class, where the word library of trigger words is composed of core words (generally, verbs and nouns are many) for identifying events and confirming event class relations. Such as "war", "conflict", etc. are triggers of birth events. According to the construction of the trigger word library and the summarization and analysis of historical data, words which appear frequently in certain type of events are used as trigger words, and the trigger words are equivalent to the type of event identifications, so that a computer can identify event types and event type relations through the trigger words.
In this embodiment, specifically, step S1 further includes acquiring an event text, performing word segmentation processing and format normalization processing on the event text, and giving part-of-speech tags corresponding to all words in the text according to a grammar rule of a language to which the event text belongs. Extracting unstructured data/semi-structured data content from event texts in various fields by a natural language algorithm, firstly segmenting the texts by a word segmentation algorithm, for example, precisely cutting sentences in the texts by a jieba tool according to Chinese grammar and performing part-of-speech tagging processing on the sentences, assigning a category called part-of-speech tag to each word in a given sentence, and assigning a correct part-of-speech tag to each word according to a Chinese dictionary and a Chinese grammar structure in the tool. For example, the part-of-speech tags in the chinese sentence include nouns (nouns), verbs (verbs), adjectives (adjectives), and so on, which are well-known grammatical structures. Labeling the words obtained by segmentation according to Chinese grammar composition, for example, labeling the nouns obtained by segmentation with "n", wherein the terms are labeled according to the first-level subclass of nouns, such as the human name "nr", the place name "ns", the organization group name "nt", other special names "nz", the noun idiom "nl", and the noun morpheme "ng", and further dividing the second-level subclasses of the nouns, for example, the human name "nr" comprises the second-level subclass Chinese surname "nr 1", the Chinese name "nr 2", the Japanese human name "nrj", and the transliterated human name "nrf".
In this embodiment, specifically, when processing the event text by using the natural language algorithm, it is further required to perform element disambiguation on terms having ambiguity in the text, for example, after performing entity disambiguation on the sentence "Apple early returned of 200Billion USD in 2016", it is inferred that Apple in the sentence is Apple, not a fruit. The event elements mentioned in the sentences are combined with semantic analysis, and the real meanings of words with ambiguity in the text are distinguished through the context.
In this embodiment, specifically, since the obtained event text (e.g., a web page, news, etc.) often includes an unstructured/semi-structured text, before extracting a feature vector (e.g., an element, an attribute, a relationship, an event, etc.) specified by the text from the event text, it needs to merge, eliminate redundancy, merge, and the like, for example, in a process for a chinese text, at least all documents with different formats need to be statistically processed into a txt document, and punctuation, space, english words, and stop words need to be removed.
The purpose of extracting the text feature vector is to convert unstructured text content that cannot be processed by a computer into a feature vector form that can be processed by the computer. And generating a feature vector according to the event text, wherein each dimension of the feature vector represents a lexical item in the event text. The length of the vector formed by all terms can reach tens of thousands or even millions. Generally, the feature vector corresponding to the event text is extracted from the feature vector with such high dimension, which may contain a large amount of redundant noise, and the calculation efficiency and the effect of the classification clustering of the subsequent event ontology are seriously affected. Therefore, feature selection and feature extraction are required, a feature space with the most distinguishing and expression capability is selected from the text, dimension reduction of the feature space is achieved, or a feature vector corresponding to the event text is extracted by mapping a high-dimensional feature vector to a low-dimensional vector space through feature conversion, the event text is divided and generalized according to the existing event class in the event ontology library, and if the event ontology library does not have an event ontology model corresponding to the event text, a corresponding event ontology model is added in the event ontology library.
In step S1, the event class of the event text is divided according to the event ontology library and/or the trigger thesaurus and the thesaurus, and corresponding event elements are extracted according to the event element template corresponding to the event class of the event text, where the task of extracting event elements generally includes two subtasks of event type identification and event element filling. Firstly, judging that the text content expresses an event type, namely an event class through event type identification. The event type determines the event element template that the event represents, with different types of events having different templates. For example, the template for the birth event is { character, time, place of birth }, and the template for the war attack event is { location, time, assailant, victim, number of injuries, … }. The event element refers to key information forming an event, and the event element extraction refers to extracting a corresponding event element according to the event type template to which the event element belongs and marking a corresponding element label for the event element.
In this embodiment, in step S2, the event text is subjected to semantic analysis and semantic relations between event elements and other event elements are extracted, the event text is analyzed and labeled according to categories of preset event relations, and corresponding event relations are extracted, where the semantic relations are used for performing calculation and reasoning out main content of the event text based on event element information. The summary information of the event text and the event class relationship of the event ontology are extracted, and the summary information can be directly extracted from the event text in a structured form description. The information contained in each text can be described as a group of event elements and the association and interaction among the event elements, so that the extraction of the event elements in the text and the semantic relationship among the event elements also become the basis for understanding the meaning of the text. The information extraction can be realized by extracting semantic relations between the event elements and other event elements, representing information carried by the semantic relations, and calculating and reasoning based on the information to effectively understand semantics carried by a text. For example, event elements of a terrorist event including time, place, attacker, victim, target of attack, consequence, etc. are extracted from related news reports; event elements of an attack are extracted from the international news, including an attacker, a victim, a place, a casualty, property loss, and the like.
Event type relation extraction refers to the task of detecting and identifying semantic relations between elements in a text, and packaging and linking elements representing the same semantic relations. The event ontology pair is classified into pre-specified relationship classes by relationship classification, and the output of relationship extraction can also be defined as a ternary structure (element 1, relationship class, element 2) by abstracting the relationship, indicating that a semantic relationship of a specific class exists between the element 1 and the element 2. The semantic relation category can preset seven types of relations in ACE evaluation, and can be determined according to a related knowledge base of the field to which the event ontology belongs. The semantic relationship categories then classify the semantic relationships that exist between the event ontologies.
For the extraction of event type relationships, the invention analyzes and labels the event text according to the type of the customized event type relationships and extracts the corresponding event type relationships, and makes detailed definitions for all event type relationships, for example, the event type relationships are roughly divided into an equality relationship, an inclusion relationship, a mutual exclusion relationship, a reciprocal relationship, a parallel relationship, and or relationship, wherein the above relationships include subclass relationships defined adaptively for each field, for example, the inclusion relationships include causal relationships, composition relationships, sequential relationships, and the like, which are only exemplified herein. According to the definition of the event type relation category, an extraction rule of the event relation is formulated, and automatic event type relation extraction and category marking are realized; and correcting the automatically labeled result in a manual mode, developing discussion on labeling of all event relations, and determining a final extraction result.
In this embodiment, specifically, by analyzing historical event correlations in the military field, customizing event ontology correlations in each field, mining extracted event class relationships to have possible correlation events in a database in the existing military field, analyzing the correlation between events by using a machine learning classification model using the event class relationships and event elements extracted from the event text to obtain a probability distribution of the events, and achieving the purpose of predicting possible future development situations of the events and different effects caused by different prediction results. For example, a possible evolution path of an event ontology for national debt yield uplink in an event library in an economic module is analyzed through a Bayesian network, a text feature vector is input into the Bayesian network trained in advance to obtain a prediction result of an event, the reason/event node of the event ontology for national debt yield uplink is analyzed through a plurality of classifiers from different angles, the prediction results of all the classifiers are integrated through mechanisms such as voting, and the evolution path of the event can be accurately predicted. For the fundamental and capital aspects, the evolution path of the maximum probability of "national debt profitability up" is: the method comprises the steps of M2 increasing rate in the same proportion, rising inflation higher than expectation, currency policy tightening, deposit reserve rate increasing, causing liquidity tension, capital interest rate ascending, national bond yield ascending, visualizing the above evolved paths, and the investors can pay attention to the currency policy issued by the central bank and can avoid the linkage influence caused by the events occurring later by comparing the indexes in the same period.
In this embodiment, specifically, many existing data relationships are represented by using a text-text combination method, such as map data, or personal relationship information. The traditional relational database management system such as RDBMS is not suitable for expressing the data relation among the data, and because each event ontology has thousands of event class relations with other related event ontologies, the information stored in the event class relations is even larger than the information of the event text. Because the traditional relational database focuses on characterizing the internal attributes of the elements, the relationships between the elements are usually realized by using foreign keys. The explosive growth of the internet, especially the mobile internet, inherently overwhelms the traditional relational database, and coupled with the high demand for relationships from applications such as the human relational network, it can be said that the relational database is already at a premium. And the graph database is produced as a database with emphasis on describing the relationship between the data.
The invention adopts the graph database to solve the storage problem of mass data and is more suitable for the data of the native expression graph structure. Extracting unstructured data/semi-structured data contents from a constructed event ontology library through a natural language algorithm, after carrying out information merging, redundancy elimination, conflict resolution, normalization and the like on the unstructured data/semi-structured data contents, respectively storing the unstructured data contents into a database (Neo4j) and a (elastic search) search engine database, and then displaying the unstructured data contents in a graphical form by utilizing a visualization function of a third-party visualization library (Pyechart) and a Neo4j database. The information of the single event ontology and the event class relation analysis of the correlation thereof are analyzed through the graph database, the overall context of the intelligence event is shown in the form of a graph, as shown in fig. 3, in step S401, a node is established according to the event ontology model, and the event ontology in the event ontology library is represented by one node in the graph database. Then, the event element is used for performing supplementary explanation on the attribute of the node through step S402, the event class corresponding to the node is labeled through a node label, each node is directionally connected according to the event class relationship of the node, the establishment of a graph database is completed, the node attribute is used for performing supplementary explanation on the event class element of an event body, the event class relationship between the event bodies is represented through the connection between the nodes, namely the edges in the nodes, wherein the edges in the nodes can be bidirectional or unidirectional, the event classes are distinguished through preset node labels, different node combinations are divided through different node labels, but the nodes under the same label do not necessarily contain the same attribute; after the whole graph database is built, for the connection management between any two nodes, the combination of the event type relations is represented by the node paths.
As shown in fig. 3, in step S403, specifically, the correlation between nodes is analyzed through a bayesian network classification model to obtain a probability distribution in a node direction, a node path is labeled according to a maximum probability node direction, and the node, a node attribute, a node label, and a node path are visually displayed. The relevance among the events is analyzed through a machine learning classification model, the probability distribution of the events is obtained, the possibility that the future development situation venation and the event development of the events are predicted is achieved, and the possible evolution path of the event body is visually displayed through a node path. The graph database can also compare the user requirement with a trigger word library and a near word library through a personalized algorithm, or compare the similarity of the user requirement with the semantic relation in the event ontology library to judge the real search intention of the user, retrieve the event related information and the event information with the event class relation in the event ontology library, evaluate the matching degree of the obtained event information to the user requirement, and visually display the event information to the user. After the information is obtained, the prediction results such as the retrieval result, the evolution path and the like can be evaluated, and the event ontology library, the similar meaning words and the trigger word library can be modified and supplemented according to the feedback of the evaluation result. For example, when the user search is completed and the evaluation is low, a new event ontology or a new trigger word is constructed according to the sentence text or the core word input by the user, and the event class relationship, the evolution path and the synonym library are supplemented.
In the present embodiment, specifically, as shown in fig. 4, according to the type of the event in the field focused by the user, the history search result, and the satisfaction feedback, the relevant event text and the like may be recommended in the subsequent query. The personalized search algorithm based on the content judges the user relevance of the document by comparing the interest and hobbies of the user with the content similarity of the result document, and further optimizes the Bayesian network parameters and switches the classification models. And supplementing the trigger word library and the event ontology library according to the historical retrieval data of the user. The personalized search algorithm comprises a link analysis-based method, which mainly utilizes the link relation between the ontology and the ontology, assumes that the web pages clicked and visited by the user are web pages interested by the user, and iterates through the link analysis algorithm to finally calculate the preference degree of the user to each web page. The personalized search algorithm also comprises a personalized search algorithm based on collaborative filtering, which mainly uses the idea of a recommendation system based on collaborative filtering, and the method considers that the personal information of the users which can be collected is limited, so that the method not only utilizes the personal information of the users, but also utilizes the information of other users or groups similar to the users, and personalizes the search results of the current users based on the interest preferences of the user groups and similar users. The similarity between users can be calculated by the contents of interests, hobbies, historical queries, clicked texts and the like of the users.
Furthermore, the relevance among the nodes is analyzed through a Bayesian network classification model to obtain the probability distribution of the node direction, the node paths are marked according to the maximum probability node direction, and the nodes, the node attributes, the node labels and the node paths are displayed visually.
Further, the content-based personalized search algorithm judges the user relevance of the document by comparing the user interests and hobbies with the content similarity of the result document, and pushes the relevant event text.
The present invention also provides a computer-readable storage medium, in which a computer program is stored, and the computer program, when executed by a processor, implements the steps of the above-mentioned method according to the embodiment of the present invention. The system/electronic device integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow in the method according to the embodiments of the present invention may also be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of the embodiments of the method. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, U-disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-only Memory (ROM), Random Access Memory (RAM), software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is appropriately increased or decreased as required by legislation and patent practice in the jurisdiction.
The invention also provides a military intelligence analysis visualization device, which comprises a memory and a processor, wherein the memory stores at least one section of program, and the at least one section of program is executed by the processor to realize the intelligent face changing method. As an executable scheme, the intelligent face changing device may be a desktop computer, a notebook, a palm computer, a cloud server, or other computing devices. The system/electronic device may include, but is not limited to, a processor, a memory. It will be understood by those skilled in the art that the above-described constituent structures of the system/electronic device are only examples of the system/electronic device, and do not constitute a limitation on the system/electronic device, and may include more or less components than those described above, or some of the components may be combined, or different components may be included. For example, the system/electronic device may further include an input/output device, a network access device, a bus, and the like, which is not limited in this embodiment of the present invention. The reader should understand that in the description of this specification, reference to the description of the terms "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described method embodiments are merely illustrative, and for example, the division of steps into only one logical functional division may be implemented in practice in another way, for example, multiple steps may be combined or integrated into another step, or some features may be omitted, or not implemented.
While the invention has been described with reference to specific embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (12)
1. A military intelligence analysis visualization method is characterized by comprising the following steps:
s1, obtaining an event text of a selected military field, preprocessing the event text, dividing the event class of the event text according to a pre-established event ontology base and extracting event elements of the event text;
s2, extracting the main content and the event type relation of the event text according to the event elements to obtain an event training set;
s3, inputting the event training set into a pre-trained Bayesian network classification model to obtain an event distribution probability, and predicting and deducing an event evolution path according to the event distribution probability;
and S4, visually displaying the event class relationship and the event evolution path.
2. The military intelligence analysis visualization method of claim 1, wherein: the event ontology library is pre-established by the following steps:
s101, acquiring a corresponding event text, performing part-of-speech tagging on content in the event text, and constructing a corresponding corpus;
s102, extracting event elements, event types and event type relations in an event text according to the corresponding corpus, and constructing a corresponding event ontology model of the military field;
s103, summarizing the event ontology models in the military affair field to obtain the event ontology library, and constructing a corresponding near-sense word library and a trigger word library according to the event class.
3. The method of military intelligence analysis visualization of claim 2, wherein: the event elements in S102 are ontologies constituting an event text, and the event elements in S102 include a time element T, an environment element V, an object element O, a process state element P, a language expression element L, and an action element a; the event class is used for representing an event type; the event class relationship is used for representing the event relationship among the events.
4. The method of military intelligence analysis visualization of claim 3, wherein: acquiring an event text in S1, and preprocessing the event text includes: and performing word segmentation processing and format normalization processing on the event text, and giving corresponding part-of-speech marks to all words in the text according to grammar rules of the language to which the event text belongs.
5. The method of military intelligence analysis visualization of claim 4, wherein: the S1 specifically includes: and dividing the event class of the event text according to the event ontology library and/or the trigger word library and the word library, and extracting corresponding event elements according to event element templates corresponding to the event class of the event text.
6. The method of military intelligence analysis visualization of claim 5, wherein: the S2 is specifically configured to perform semantic analysis on the event text and extract semantic relationships between event elements and other event elements, analyze and label the event text according to the type of the customized event type relationship and extract corresponding event type relationships, where the semantic relationships are used to perform calculation and inference based on the event element information to obtain the main content of the event text.
7. The method for analyzing and visualizing military intelligence of claim 6, wherein the step of S4 specifically comprises the steps of:
s401, establishing nodes according to the event ontology model;
s402, performing supplementary description on the attribute of the node according to the event element, marking an event class corresponding to the node through a node label, and connecting each node according to the event class relation of the node to complete the establishment of a graph database;
and S403, analyzing the correlation between the nodes through the Bayesian network classification model to obtain the probability distribution of the node direction, marking the node path according to the maximum probability node direction, and visually displaying the nodes, the node attributes, the node labels and the node path.
8. The military intelligence analysis visualization method according to claim 7, wherein all the nodes are visually displayed according to event texts and/or trigger words input by users and synonym words for searching the relevant nodes and the nodes having event-like relations with the relevant nodes.
9. The method of military intelligence analysis visualization of claim 8, wherein the event-class relationships comprise geospatial relationships, and wherein all of the nodes are visualized based on geospatial and provide corresponding data manipulation portals.
10. The military intelligence analysis visualization method of claim 9, wherein the user interest is judged by comparing the similarity of the event class and the search result in the user search history according to a personalized search algorithm, and then the search results are ranked.
11. A computer readable storage medium having stored thereon at least one program for execution by a processor to implement the method for visualization of military intelligence analysis of any of claims 1-10.
12. A military intelligence analysis visualization apparatus comprising a memory and a processor, wherein the memory stores at least one program, and the at least one program is executed by the processor to implement the military intelligence analysis visualization method of any of claims 1 to 10.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210054358.7A CN114896387A (en) | 2022-01-18 | 2022-01-18 | Military intelligence analysis visualization method and device and computer readable storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210054358.7A CN114896387A (en) | 2022-01-18 | 2022-01-18 | Military intelligence analysis visualization method and device and computer readable storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114896387A true CN114896387A (en) | 2022-08-12 |
Family
ID=82714615
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210054358.7A Pending CN114896387A (en) | 2022-01-18 | 2022-01-18 | Military intelligence analysis visualization method and device and computer readable storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114896387A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115907144A (en) * | 2022-11-21 | 2023-04-04 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Event prediction method and device, terminal equipment and storage medium |
CN116821374A (en) * | 2023-07-27 | 2023-09-29 | 中国人民解放军陆军工程大学 | Event prediction method based on information |
-
2022
- 2022-01-18 CN CN202210054358.7A patent/CN114896387A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115907144A (en) * | 2022-11-21 | 2023-04-04 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Event prediction method and device, terminal equipment and storage medium |
CN116821374A (en) * | 2023-07-27 | 2023-09-29 | 中国人民解放军陆军工程大学 | Event prediction method based on information |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ittoo et al. | Text analytics in industry: Challenges, desiderata and trends | |
Ding et al. | Applications of natural language processing in construction | |
Sun et al. | A review of natural language processing techniques for opinion mining systems | |
Xia et al. | Dual sentiment analysis: Considering two sides of one review | |
Gao et al. | A rule-based approach to emotion cause detection for Chinese micro-blogs | |
Bordoloi et al. | Sentiment analysis: A survey on design framework, applications and future scopes | |
Zubrinic et al. | The automatic creation of concept maps from documents written using morphologically rich languages | |
Athar | Sentiment analysis of scientific citations | |
Mohamed et al. | A hybrid approach for paraphrase identification based on knowledge-enriched semantic heuristics | |
Kmail et al. | An automatic online recruitment system based on exploiting multiple semantic resources and concept-relatedness measures | |
Zou et al. | Learning to rank for question-oriented software text retrieval (t) | |
US11893537B2 (en) | Linguistic analysis of seed documents and peer groups | |
CN112989208B (en) | Information recommendation method and device, electronic equipment and storage medium | |
Khademi et al. | Persian automatic text summarization based on named entity recognition | |
Ali et al. | Semantic similarity measures between words: A brief survey | |
CN114265936A (en) | Method for realizing text mining of science and technology project | |
CN114896387A (en) | Military intelligence analysis visualization method and device and computer readable storage medium | |
Tovar et al. | Identification of Ontological Relations in Domain Corpus Using Formal Concept Analysis. | |
Rybak et al. | Machine learning-enhanced text mining as a support tool for research on climate change: theoretical and technical considerations | |
Vanetik et al. | Multilingual text analysis: History, tasks, and challenges | |
Rybina | Sentiment analysis of contexts around query terms in documents | |
Arumugam | Processing the textual information using open natural language processing (NLP) | |
Akkineni et al. | Hybrid Method for Framing Abstractive Summaries of Tweets. | |
da Costa | Semantic Enrichment of Knowledge Sources Supported by Domain Ontologies | |
Gao | A combined rule-based and machine learning approach for blackout analysis using natural language processing |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |