CN113268591B - Air target intention evidence judging method and system based on affair atlas - Google Patents

Abstract

The invention belongs to the technical field of air target intention judgment, and particularly relates to an air target intention judgment evidence method and system based on a case atlas, wherein typical event types in the activity field of the air target are obtained according to air target flight activity text data; setting event trigger words and event elements corresponding to each type of typical events, analyzing text data of the flight activities of the aerial targets to extract the events, and acquiring typical events and atypical events in the flight activities; marking label attributes on all events and combining a database to construct an air target flight activity event map; and aiming at the flight activity data of the air target to be processed, judging and verifying the intention of the air target by extracting the event and matching the event type based on the event map. The invention carries out intention judgment by means of the constructed case map, promotes the comprehensive grasp of the practical application environment situation of the aerial target, is beneficial to the timely management and control of the aerial target and has better application prospect.

Description

Method and system for judging air target intention based on event map

technical field

The invention belongs to the technical field of aerial target intention discrimination, and in particular relates to a method and system for aerial target intention discrimination based on an event graph.

Background technique

With the continuous development of modern aircraft technology, the types of various application tasks are increasing. In practical applications, the timely and accurate identification of high-value aerial target intentions can ensure the overall control of the application environment situation, and is conducive to real-time analysis and judgment of the application environment to make timely responses. For both application parties, the determination of the intention of air targets is complicated and requires a lot of time and manpower. In the face of complicated flight activity text data of air targets, there is a relatively obvious error rate in manual judgment of intentions, so it is urgent to develop an intention judgment method that can assist personnel in the judgment and has better results.

The current research on judging the intention of air targets mainly adopts two methods: one is multi-attribute decision-making, and the other is knowledge-based reasoning. Multi-attribute decision-making can comprehensively process a large amount of information but lacks reasoning ability, and the method based on knowledge reasoning is too computationally intensive and its completeness is difficult to guarantee. For example, air target intent recognition based on deep neural network is an improvement to multi-attribute decision-making. The rules between feature states and intent are obtained through self-training, which is used to represent the corresponding relationship between the two and improve the accuracy of identifying intent. As another example, the air target intention discriminant model based on intuitionistic fuzzy production rule reasoning (IPR) and multi-attribute decision-making combines the advantages of multi-attribute decision-making and knowledge reasoning to improve the accuracy of the results, but the specific feasibility is still unclear. to be explored.

Contents of the invention

To this end, the present invention provides a method and system for judging the intention of an air target based on an event map. Through the existing flight activity text data of an air target, the flight activity events are extracted, and the evolution rules and patterns between the events are mined. Construct an air target intention and event map, and use the constructed event map to conduct intent judgment, improve the overall grasp of the actual application environment situation of air targets, and facilitate timely management and control of air targets.

According to the design scheme provided by the present invention, a method for judging the intention of an air target based on an event map is provided, which includes the following contents:

Obtain the typical event types in the field of activity of the air target based on the text data of the flight activity;

By setting the event trigger words and event elements corresponding to each type of typical event, and analyzing the flight activity text data of the air target to perform event extraction, obtain typical events and atypical events in its flight activities;

By labeling label attributes for all events and combining with graph databases, an event map of air target flight activities is constructed;

For the flight activity data of the air target to be processed, the event type is matched through event extraction and based on the event map, and the air target intention is judged.

As the air target intention judgment method based on the event map of the present invention, further, for the air target flight activity text data, firstly perform character replacement and data cleaning through regular expressions, and then obtain typical events according to the word clustering method.

As the air target intention judgment method based on the event map of the present invention, further, in the acquisition of typical event types, by performing word segmentation, part-of-speech tagging and grammatical analysis on the text data, the subject-predicate relationship and the verb-object relationship binary group are extracted to determine the flight Candidate trigger words for activity events form a set of candidate trigger words; filter the trigger words in the set of candidate trigger words through preset filtering rules, and perform lexical clustering based on the semantic similarity of words to form typical event types in the field of flight activities.

As the method for judging the intention of aerial targets based on the event map of the present invention, further, the preset filtering rules include but are not limited to: retain the general verbs and gerunds in the set of candidate trigger words, and filter out other types of verbs; and/or, according to the set The degree of correlation between the candidate trigger words and the flight activity field of the air target, and filter out the trigger words whose correlation degree is less than the set threshold.

As the air target intention judgment method based on the event map of the present invention, further, the typical event types obtained include: takeoff, accompanying flight, stopover, arrival, ordinary flight, deployment, circling and landing.

As the method for judging the intention of the air target based on the event map of the present invention, further, in the event extraction, the event type is identified by analyzing the flight activity text data of the air target, and if it is a typical event type, the event is extracted by named entity and pattern matching ; If it is an atypical event type, the atypical event trigger words and event elements are obtained by analyzing the relationship between syntactic components.

As the method for judging the intention of the aerial target based on the event map of the present invention, further, for atypical event types, according to the analysis results of the relationship between the syntactic components, the core word is set as the event trigger word, and combined with the attributive acquisition of the subject and object components event element.

As the air target intention judgment method based on the event map of the present invention, further, in the construction of the event map, all events are uniquely numbered, the relationship between the flight activity events of the air target is excavated, and multiple events in the same data are marked through tag attributes. Events with the same label attribute are set to have a sequential relationship, and the graph database is used to construct and store the flight activity event map of air targets.

As the air target intention judgment method based on the event map of the present invention, further, for the flight activity data of the air target to be processed, firstly perform event extraction, and match the event type through the event map; if the event type cannot be matched through the event map, Then use keywords to carry out fuzzy matching to determine the affiliation relationship of event types.

Further, the present invention also provides an air target intention judgment system based on the event map, including: a type acquisition module, an event extraction module, a map construction module and an intention matching module, wherein,

The type acquisition module is used to obtain the typical event type in its activity field according to the flight activity text data of the air target;

The event extraction module is used to perform event extraction by setting the event trigger words and event elements corresponding to each type of typical event, and analyzing the flight activity text data of air targets to obtain typical and atypical events in its flight activities. event;

The map building module is used to construct an event map of air target flight activities by annotating tag attributes for all events and combining with a graph database;

The intention matching module is used for judging the intention of the air target by extracting events and matching event types based on the event graph for the flight activity data of the air target to be processed.

Beneficial effects of the present invention:

The present invention discovers typical events in this field starting from the flight activity text data of the air target, performs structural representation on the unstructured text information, and conducts target intention discrimination by constructing the flight activity event map of the air target and the air target intention event map ;Adopt the method of field event word clustering to discover typical events in the field of air target flight activities, so that the definition of event types and event argument roles in the field of air target flight activities is more clear; the constructed air target flight activity map and intention event map are intuitive Clearly displaying the changes in flight activities of air targets and the relationship between flight activities and intentions is conducive to the identification of air targets' intentions and the real-time monitoring of air targets, and has a good application prospect.

Description of drawings:

Fig. 1 is a schematic flow chart of the air target intention judgment method based on the matter map in the embodiment;

FIG. 2 is a schematic diagram of a typical event type clustering process in an embodiment;

FIG. 3 is a schematic diagram of an event extraction process in an embodiment;

Fig. 4 is a schematic diagram of the dependency syntax analysis structure in the embodiment;

Fig. 5 is a schematic diagram of the target intention event map in the embodiment.

Detailed ways:

In order to make the purpose, technical solution and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and technical solutions.

An embodiment of the present invention provides a method for judging the intention of an air target based on an event map, which includes the following content: obtain the typical event types in the field of activity of the air target according to the flight activity text data; event trigger words and event elements, and analyze the flight activity text data of air targets to extract events, obtain typical events and atypical events in their flight activities; construct airborne Target flight activity event map: for the air target flight activity data to be processed, events are extracted and event types are matched based on the event map to determine the air target intention.

Through the existing flight activity text data of air targets, the flight activity events are extracted, the evolution rules and patterns between events are mined, and the air target intention and reason map is constructed, and the intention is judged with the help of the constructed matter map. The rationality of judging the intention of air targets by constructing a matter map lies in: first, the flight activities of air targets have certain rules and patterns, and the matter map can just describe the flight activities of air targets; second, the edges of the matter map represent The relationship between events, and the sequential relationship and causal relationship. The event map is suitable for the study of the intention of the flight target: the flight activity is regarded as a series of events with a sequential relationship, and this series of events can be abstracted into a reason Events, the intention as the result event, the causal relationship between the two as the causal relationship between events in the event map. In terms of application, mining the succession relationship of air target flight activities and the causal relationship between air target flight activities and intentions, and visually displaying them in the form of event maps can be applied to the causal analysis of air target flight activities and intentions. The intention judgment of air targets provides new analytical tools and approaches, provides support for predicting intentions, and even assists in predicting changes in flight activities of air targets. Use the "Affair Map" to judge the intention of the air target. From a new perspective, use the advantages of the event map to display the flight activities of the air target in the form of a map, assist the judgment of intention, and improve the accuracy of personnel judgment.

In the field of air target flight activities, the definition of event types and event argument roles is not clear, the amount of air target flight activity message data obtained from open sources is small, and the scale of the constructed event graph is small, which leads to unsatisfactory target intention judgment results, etc. , in the embodiment of this case, as shown in Figure 1, the typical events in this field are discovered based on the clustering of event words in the field of air target flight activities, and event templates are designed for each event; The problem obtains the target intention through the new keyword fuzzy matching method.

As an air target intention judgment method based on the event map in the embodiment of the present invention, further, for the air target flight activity text data, first perform character replacement and data cleaning through regular expressions, and then obtain typical events according to the word clustering method.

Use regular expressions to replace characters, clean the text data of air target flight activities, and use domain event word clustering to discover typical events in the field of air target flight activities. The specific steps include word segmentation, part-of-speech tagging, dependency syntax analysis, and verbs. Subdivision, domain correlation calculation, semantic similarity calculation, etc. Further, in the acquisition of typical event types, through word segmentation, part-of-speech tagging and grammatical analysis of text data, subject-predicate relationship and verb-object relationship binary groups are extracted, and candidate trigger words for flight event events are determined to form a set of candidate trigger words; Filtering rules are set to filter the trigger words in the set of candidate trigger words, and vocabulary clustering is performed based on the semantic similarity of words to form typical event types in the field of flight activities. As shown in Figure 2, word segmentation, part-of-speech tagging and dependency syntactic analysis are performed on the flight activity text of the air target, and the subject-predicate SBV and verb-object VOB pairs are extracted, and words such as the verb "take off" that serve as important components in the sentence are determined as The candidate trigger words for flight activity events form a set of candidate trigger words. Aiming at the noise data in the set of candidate trigger words, two filtering rules of retaining general verbs, gerunds and domain relevance in the candidate trigger words are used to filter the trigger words. The preset filtering rules may include but are not limited to: keep general verbs and gerunds in the set of candidate trigger words, and filter out other types of verbs; and/or, filter out related Trigger words whose degree is less than the set threshold. For example, the filter rule for retaining general verbs and gerunds in candidate trigger words is to divide verbs into linking verbs VX, auxiliary verbs VZ, formal verbs VF, directional verbs VQ, complementary verbs VB, general verbs VG, noun verbs VN, and adverbs There are eight categories of VD; the domain correlation filter rule is the domain correlation to reflect the degree of correlation between the candidate trigger word and the flight activity domain of the air target, using (the frequency of the candidate trigger word in V’s expected appearance in the domain)/(the candidate trigger word in frequency of occurrence in the general domain) calculations. For the similarity between the meaning and usage of trigger words, the word semantic similarity calculation based on "Synonyms Ci Lin (Extended Edition)", the word semantic similarity calculation based on HowNet's original concept of word semantics and the word based on Word2Vec The semantic similarity algorithm can cluster words with high similarity to complete event-triggered word clustering. Typical event types obtained after trigger word extraction, filtering, and clustering include: takeoff, accompanying flight, stopover, arrival, ordinary flight, deployment, circling, and landing.

As an air target intention judgment method based on the event map in the embodiment of the present invention, further, in event extraction, the event type is identified by analyzing the flight activity text data of the air target, and if it is a typical event type, it is identified by named entity and pattern matching. Perform event extraction; if it is an atypical event type, obtain atypical event trigger words and event elements through relationship analysis between syntactic components. Furthermore, for atypical event types, according to the analysis results of the relationship between syntactic components, the core word is set as the event trigger word, and the event elements are obtained by combining the attributives of the subject and object components.

Analyze the text data of air target flight activities to identify the type of event. For those identified as typical events, the method of named entity recognition combined with pattern matching is used for event extraction, and for those identified as atypical events, event extraction is performed based on dependency syntax analysis. Among them, typical event extraction mainly includes the extraction of air target flight time, location and entity (type, call sign, number and quantity, etc.). As shown in Figure 3, for the identification of time and place, the method of named entity recognition combined with pattern matching is used. An accurate named entity recognition task can identify time, and then perform pattern matching in combination with the common forms of time expression, and identify places and organizations. The entity of the organization is used as the location, combined with the design mode "taking xx as the take-off and landing place", "flying over xx" and other extraction locations. For the extraction of entities, mainly model, call sign, number, quantity, etc., the pattern matching method can be used to consider the characteristics of various attributes to extract qualified letter and number strings, for example: the number is generally 6 or 7 digits , There is a horizontal bar in the middle of the model, etc. For atypical events, since there is no fixed pattern designed for the events in the text, it is impossible to determine which elements the event should have. Therefore, the method based on dependency syntax analysis is used to obtain the description in the text by analyzing the relationship between syntactic components. The main element of the event. Extract the trigger words and event elements of atypical events based on dependency syntax analysis, first set the core word as the event trigger word according to the result of dependency syntax analysis, and generally the core word is the predicate of the sentence; then extract the subject and object in the result of dependency syntax analysis , an attributive that combines these two components as an event element.

As an air target intention judgment method based on the event map in the embodiment of the present invention, further, in the construction of the event map, all events are uniquely numbered, the relationship between the air target flight activity events is excavated, and the same data is marked through the label attribute For multiple events in the system, events with the same tag attribute are set to have a sequential relationship, and a graph database is used to construct and store the flight activity event map of air targets. Further, for the flight activity data of air targets to be processed, event extraction is performed first, and the event type is matched through the event map; if the event type cannot be matched through the event map, the keyword is used for fuzzy matching to determine the event type affiliation.

Use the flight activity data of the air target to perform event extraction according to the above steps, and extract the typical events and atypical events of the air target flight activity. As shown in Figure 5, all events are uniquely numbered id, so that each event can be represented by its number, and the relationship between air target flight activity events is mined by setting specific attributes, and the "label" attribute is used to mark the data in a piece of data. For multiple events, events with the same "label" attribute have a sequential relationship, and the graph database is used to construct and store the flight activity event map of air targets. Determine the key elements of various air target flight activity events, and perform operations such as node merging, intention association, and confidence assignment on the air target flight activity event map to construct the air target intention event map. And based on the air target intention map and keyword fuzzy matching method, the air target intention is judged. First, event extraction is performed to match and identify elements such as event type, model, and location in the extracted structured events. Cypher query statements can be written, and the generated Cypher statements can be used to match intents in the intent event map. When the intention cannot be matched by using the intention event map, the intent matching method is changed to keyword fuzzy matching to determine the affiliation relationship between the event element keyword, type, location and intention.

Further, based on the above-mentioned method, the embodiment of the present invention also provides an air target intention judgment system based on the event map, including: a type acquisition module, an event extraction module, a map construction module and an intention matching module, wherein,

The type acquisition module is used to obtain the typical event type in its activity field according to the flight activity text data of the air target;

The event extraction module is used to perform event extraction by setting the event trigger words and event elements corresponding to each type of typical event, and analyzing the flight activity text data of air targets to obtain typical and atypical events in its flight activities. event;

The map building module is used to construct an event map of air target flight activities by annotating tag attributes for all events and combining with a graph database;

The intention matching module is used for judging the intention of the air target by extracting events and matching event types based on the event graph for the flight activity data of the air target to be processed.

In order to verify the effectiveness of the scheme in this case, the following is a further explanation based on specific experimental data:

The air target flight activity message data is stored in a .txt file, as follows:

On the morning of February 29, a KC-10A tanker (MOJO91) numbered 83-0081 of an air force took off from the base A11 in area A and flew over area B.

On the evening of January 30, an RC-135U reconnaissance aircraft (GIVE31) of an air force with the serial number 64-14847 took off from C11 base in area C and flew over area D.

On January 27, a MC-130J special aircraft (RCH1031) numbered 08-6206 of the 27th Special Wing of an Air Force took off from base E, stopped at areas E1, E2, and E3, and finally landed at base F.

On the afternoon of January 30, an EP-3E reconnaissance aircraft (MN806) numbered 156517 of a navy took off from area G and flew over area H.

On the morning of January 30, an RC-135V reconnaissance aircraft (PYTHN54) numbered 64-14848 of an air force lifted off from base I and flew over area H.

...

The first step is to use the domain event word clustering method to discover typical events in the field of air target flight activities, taking "a B-52H bomber takes off from base L in city K in area J" as an example.

1a) Word segmentation and part-of-speech tagging results for the flight activity text of the air target: 1 (m), frame (q), B-52H (n), bomber (n), from (p), J area (n), K city (n), base L (n), takeoff (v).

The results of dependency syntax analysis are as follows:

Component relationship element of sentence ATT (1, frame) ATT (frame, bomber) ATT (B-52H, bomber) SBV (bomber, take off) ADV (from, take off) ATT (J area, K city) ATT (K city, base L) POB (base L, from) HEAD (take off)

Extract subject-predicate SBV and verb-object VOB dyads from it, only (bomber, takeoff) meets the conditions, and finally extract "takeoff" in which the part of speech is a verb, as a candidate trigger word. Use this method to extract trigger words from a large amount of text data to obtain a set of candidate trigger words.

1b) Filter the noise data in the set of candidate trigger words.

Filtering rule 1: keep general verbs and gerunds in the extracted candidate trigger words, and discard other types of verbs.

Filtering rule 2: Use domain correlation to reflect the degree of correlation between the candidate trigger word and the flight activity domain of the air target. The calculation formula is as follows:

DR(V)=(Freq_p(V))/(Freq_G(V))

Among them, DR(V) is the domain correlation value of the candidate trigger word V, Freq_p(V) is the frequency of the candidate trigger word V appearing in the domain corpus, and Freq_G(V) is the frequency of the candidate trigger word V appearing in the general domain corpus frequency. After sorting, select the top word as the final trigger word.

1c) Use semantic similarity to cluster event-triggered words, and get words with high semantic similarity such as "take off" and "launch"; "stop", "dock" and "stop"; "fly over", "fly " and "taxi"; "deployment", "deployment" and "arming" and so on.

The second step is to design trigger words and event elements corresponding to each type of typical event type, as follows.

The third step is to analyze the text data of the flight activity of the air target to identify the event type.

3a) Typical event extraction mainly includes the extraction of air target flight time, location and fighter entity (fighter model, fighter call sign, fighter number and quantity). Take "On the morning of February 29, a KC-10A tanker (MOJO91) numbered 83-0081 of an air force took off from the base A11 in area A and flew over area B." For example, the extracted events are:

Event 1: {'events':'Takeoff','trigger':'Takeoff','time':'The morning of February 29','location': ['A region','Base A11'],'Model ':['KC-10A'],'call sign':['MOJO91'],'number':['83-0081'],'quantity':['1']}

Event 2: {'events':'Normal flight','trigger':'Flight','time':'The morning of February 29th','location':['Area B'],'model':[ 'KC-10A'],'call sign':['MOJO91'],'number': ['83-0081'],'quantity':['1']}

3b) Atypical event extraction, by analyzing the relationship between syntactic components, to obtain the main elements describing events in the text. The existing Chinese text dependency syntax analysis has a total of 15 annotation relations, namely subject-predicate relationship (SBV), verb-object relationship (VOB), inter-object relationship (IOB), pre-object (FOB), double language (DBL), ATT, ADV, CMP, COO, POB, LAD, RAD, independent structure (IS), Punctuation (WP), and Core Relations (HED). For example: "A C-17 transport aircraft has entered the airspace of area B1". The structure of dependency syntax analysis is shown in Figure 4:

The core word is "entry"; the subject is "transport plane"; the predicate is "airspace"; the subject and predicate are combined with the attributive component (ATT) to obtain the event elements as "1 C-17 transport plane" and "area B1 airspace". The event finally extracted is {(1 C-17 transport aircraft); (entry); (area B1 airspace)}.

In the fourth step, due to the particularity of the flight activity data of the air target, the data describes the flight activity of the air target by line, and separates multiple flight activities of an air target by commas to obtain a field containing a single event. The data of a typical event has 14 attributes, which are: event type, trigger word, time, model, call sign, number, quantity, model 2, call sign 2, number 2, quantity 2, direction, location and label. Among them, the four attributes of model 2, call sign 2, serial number 2, and quantity 2 are for the storage of "accompanying object" in "accompanied flight event". If it is not "accompanied flight event", this attribute is judged to be empty, and the direction attribute is for "common flight event". The storage of the flight direction in the middle, if it is not a "common flight event", this attribute is judged to be empty. The label attribute has the function of identification, and marks multiple events of one flight activity of the air target. This paper uses numbers to mark. Atypical event data has four attributes, namely: event type, trigger word, event element, and label. The label is also used to label multiple events of the same flight activity. Finally, 341 typical events and 21 atypical events were extracted. Events, including some abnormal events.

The fifth step is to construct an air target flight activity event map according to the event number id, label and relationship, and aggregate events with the same "label" attribute in sequence to obtain multiple (id1, id2) binary groups.

The sixth step is to perform operations such as node merging, intention association, and confidence assignment on the air target flight activity event map to construct the air target intention event map.

6a) The merging of event nodes is mainly aimed at filtering the secondary attributes of events. The intention of an air target is generally related to the model of the fighter and the location of the fighter’s activities, while attributes such as time, call sign, serial number, and quantity play little role. When merging event nodes, only the event type, fighter model, and activity location are kept. Attributes.

6b) Intent association, creating an intention node, and then matching the flight activity event node and the intention node according to the existing relationship data between the flight activity event and the intention, and creating the relationship.

6c) The calculation of the confidence includes the calculation of the confidence degree of the succession relationship between the flight activity event nodes and the calculation of the causal relationship confidence degree between the flight activity event nodes and the intention. The calculation method of the confidence degree of the succession relationship mainly adopts the method of counting. According to the event id relationship binary group obtained in the previous section, the same event node is counted pointing to other event nodes, and then the proportion is calculated separately to obtain the weight value, which is the confidence Spend. Confidence in causality is similar.

In the seventh step, the intent is first judged based on the constructed intention-matter graph, but considering the small scale of the graph, there may be cases where nodes cannot be matched. If the intent node cannot be matched, the intent will be matched using keyword fuzzy matching.

7a) Carry out intention discrimination based on the air target intention incident map, match and identify the event type, model, location and other elements in the extracted structured events according to specific regular expressions, write Cypher query sentences, and use the generated Cypher sentences in Match intents in the intent event graph. The Cypher statement matched with the event elements extracted in 3a) is:

MATCH(a: takeoff{type:'KC-10A'})-[r1:success]->(b:ordinary flight{type:'KC-10A'})-[r2:cause and effect]->(c: INTENT) WHERE(a.location CONTAINS 'Area A, base A11' OR 'Area A, base A11' CONTAINS a.location) AND (b.location CONTAINS 'over area B' OR 'over area B'CONTAINSb.location) RETURN (c.type)

The matching result in the air target intention and event map is "refueling".

7b) Due to the small scale of the affair map, there are cases where the intention cannot be matched. Intent matching is carried out in the way of keyword fuzzy matching. The core of keyword fuzzy matching intent is to determine the affiliation relationship between certain event element keywords and intentions based on fuzzy statistics based on experience and human psychological processes. Combining the experience of relevant personnel and the research on text data, the affiliation relationship between event element keywords and intentions as shown in Table 1, Table 2 and Table 3 is obtained:

Table 1. Event type and intent affiliation table

Event Type Keywords intention deploy military deployment, military support circling reconnaissance escort Come on, pilot, come on

Table 2 Entity type and intent affiliation table

entity type intention transport plane Transport personnel, transport materials tanker refueling, refueling pilot reconnaissance aircraft reconnaissance training machine training, drill fighter deterrence special mission aircraft reconnaissance

Table 3 Location and intent affiliation table

Take "on the evening of January 30, an RC-135U reconnaissance aircraft (GIVE31) of a certain air force with the number 64-14847 lifted off from C11 base in area C and flew over area D." as an example. Assuming that it cannot match the intention on the map of air target flight activities, extract its key elements: "reconnaissance aircraft" and "over the Persian Gulf", and according to the affiliation table, it can be obtained that its intention is "reconnaissance of Iran".

Another example is to obtain target activity text data of aerial vehicles such as drones, and judge its true intention (simulation training, performance display or others) based on the text data. By adopting the method of field event word clustering to discover typical events in the field of air target flight activities, the definition of event types and event argument roles in the field of air target flight activities is more clear; and using the constructed air target flight activity map and intention events The map intuitively and clearly shows the changes in the flight activities of air targets and the relationship between flight activities and intentions, which is conducive to the identification of the intentions of air targets, facilitates drills, simulation training and the application of virtual reality technology in games, and enhances participants Real-time interactive perception effect.

Relative steps, numerical expressions and numerical values of components and steps set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

It should be noted that like numerals and letters denote similar items in the following figures, therefore, once an item is defined in one figure, it does not require further definition and explanation in subsequent figures.

Finally, it should be noted that: the above-described embodiments are only specific implementations of the present invention, used to illustrate the technical solutions of the present invention, rather than limiting them, and the scope of protection of the present invention is not limited thereto, although referring to the foregoing The embodiment has described the present invention in detail, and those skilled in the art should understand that any person familiar with the technical field can still modify the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention Changes can be easily thought of, or equivalent replacements are made to some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of the present invention within the scope of protection. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (10)

1. A method for judging the intent of an aerial target based on the map of affairs, characterized in that it comprises the following content: Obtain the typical event types in the field of activity based on the text data of the flight activity of the air target; By setting the event trigger words and event elements corresponding to each type of typical event, and analyzing the flight activity text data of the air target to perform event extraction, obtain typical events and atypical events in its flight activities; Construct an air target flight activity event map by marking all events with label attributes and combining with graph database; determine the key elements of various air target flight event events, and perform node merging, intention association and confidence assignment on the air target flight event event map operate to build an airborne target intent and reason map; For the air target flight activity data to be processed, the event type is matched through event extraction and based on the air target intention event map, and the air target intention is judged. 2. the method for judging the intention of the aerial target based on the matter map according to claim 1, characterized in that, for the text data of the aerial target flight activity, at first, character replacement and data cleaning are carried out by regular expressions, and then according to the word clustering method Get typical events. 3. according to claim 1 or 2 described method based on the air target intention discrimination method of event map, it is characterized in that, in typical event type acquisition, by carrying out word segmentation, part-of-speech tagging and grammatical analysis to text data, extract subject-predicate relationship and Verb-object relationship binary group, determine the candidate trigger words for flight activity events, and form a candidate trigger word set; filter the trigger words in the candidate trigger word set through preset filtering rules, and perform lexical clustering based on the semantic similarity of words to form Typical event types in the field of flight activities. 4. The method for judging the intention of an aerial target based on an event map according to claim 3, wherein the preset filtering rules include but are not limited to: retain general verbs and gerunds in the set of candidate trigger words, and filter out other types of verbs and/or, according to the degree of correlation between the candidate trigger words in the set and the flight activity field of the air target, filter out the trigger words whose correlation degree is less than the set threshold. 5. The air target intention judgment method based on the event map according to claim 3, wherein the acquired typical event types include: takeoff, accompanying flight, stopover, arrival, ordinary flight, deployment, circling and landing. 6. The method for judging the intention of an air target based on an event map according to claim 1, wherein, in event extraction, the event type is identified by analyzing the text data of the flight activity of the air target, and if it is a typical event type, then by naming the entity and pattern matching for event extraction; if it is an atypical event type, the atypical event trigger words and event elements are obtained by analyzing the relationship between syntactic components. 7. the air target intention judgment method based on the matter map according to claim 6, is characterized in that, for atypical event types, according to the relationship analysis results between syntactic components, core words are set as event trigger words, and combined The attributives for the subject and object components capture the event elements. 8. the air target intention judgment method based on the event map according to claim 1 is characterized in that, in the construction of the air target flight activity event map, all events are uniquely numbered, and the relationship between the air target flight event is excavated , and label multiple events in the same data through label attributes, set the sequence relationship between events with the same label attributes, and use the graph database to construct and store the flight activity event map of air targets. 9. according to claim 1 or 8 described based on the air target intention discriminating method of matter map, it is characterized in that, for the air target flight activity data to be processed, at first carry out event extraction, and match event by air target intention matter map type; if the event type cannot be matched through the air target intention incident map, then use keywords to perform fuzzy matching to determine the event type affiliation. 10. A system for judging the intention of an air target based on an incident map, characterized in that it includes: a type acquisition module, an event extraction module, a map construction module and an intention matching module, wherein, The type acquisition module is used to obtain the typical event type in its activity field according to the text data of the flight activity of the air target; The event extraction module is used to perform event extraction by setting the event trigger words and event elements corresponding to each type of typical event, and analyzing the flight activity text data of air targets to obtain typical and atypical events in its flight activities. event; The map construction module is used to construct the air target flight activity event map by marking all events with tag attributes and combining with the graph database; determine the key elements of various air target flight event events, and perform node merging on the air target flight event event map, Intent association and confidence-assignment operations to construct an air target intent and reason map; The intent matching module is used to match the event type through event extraction and match the event type based on the air target intention event map for the air target flight activity data to be processed, and judge the air target intent.

Images