CN112925899B - Ordering model establishment method, case clue recommendation method, device and medium - Google Patents
Ordering model establishment method, case clue recommendation method, device and medium Download PDFInfo
- Publication number
- CN112925899B CN112925899B CN202110181900.0A CN202110181900A CN112925899B CN 112925899 B CN112925899 B CN 112925899B CN 202110181900 A CN202110181900 A CN 202110181900A CN 112925899 B CN112925899 B CN 112925899B
- Authority
- CN
- China
- Prior art keywords
- data
- case
- management objects
- recommendation
- relationship
- 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.)
- Active
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
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
-
- 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
- G06F16/33—Querying
- G06F16/338—Presentation of query results
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention relates to the technical field of clue recommendation, in particular to a method for establishing a sequencing model, a method, a device and a medium for recommending case clues. The invention aims to solve the problems of single logic and poor accuracy of the traditional case clue recommendation system. For this purpose, the model building method of the present invention comprises: acquiring multi-modal data related to a plurality of management objects; determining relationship data between a plurality of management objects based on the multimodal data; generating labels related to a plurality of management objects according to the multi-modal data and the relationship data; generating a joint representation based on the relationship data and the labels; based on the joint representation, a ranking model is established for ranking the plurality of management objects. The multi-mode data of the management object is acquired for learning modeling, the multi-mode mass data of the service party is fully utilized, the description of the management object is more stereoscopic, and the recommended accuracy and coverage rate can be greatly improved.
Description
Technical Field
The invention relates to the technical field of clue recommendation, in particular to a method for establishing a sequencing model, a method, a device and a medium for recommending case clues.
Background
The case clue recommending system is a system for automatically recommending management objects possibly related to a case for criminal investigation personnel based on mining of massive data when case analysis is performed on criminal investigation scenes. Wherein the management object comprises people, things, places, things, organizations and the like.
With the development of big data technology, the current case clue recommendation system basically solves the problem of real-time recommendation of massive data. However, existing case recommendation system schemes are typically based on single modality data, such as: and constructing an entity relationship map among people, things and places through the case file data, and generating a semantic tree model for the input case to recommend. The data resources available to the service party (i.e. criminal investigation party) are multi-modal, and the existing case clue recommendation system does not fully utilize the abundant data of the service party, so that the accuracy is often poor after the system is online, or the recommendation clue logic is too single, and the service value is low.
Accordingly, there is a need in the art for a new solution to the above-mentioned problems.
Disclosure of Invention
In order to solve at least one of the above problems in the prior art, that is, to solve the problem of single logic and poor accuracy existing in the existing case thread recommendation system, in a first aspect of the present application, a method for creating a ranking model for case thread recommendation is provided, where the method includes:
Acquiring multi-modal data related to a plurality of management objects;
determining relationship data between a plurality of the management objects based on the multimodal data;
generating labels related to a plurality of management objects according to the multi-modal data and the relation data;
generating a joint representation based on the relationship data and the tag;
based on the joint representation, a ranking model is established for ranking the plurality of management objects.
In the preferred technical solution of the ranking model building method for case cue recommendation, after the step of determining the relationship data between the plurality of management objects, the method further includes:
and storing the relation data in a knowledge graph.
In the preferred technical solution of the method for creating a ranking model for case cue recommendation, the multimodal data includes: personnel static information data, dynamic track data and case file data.
In the above preferred technical solution of the ranking model building method for case cue recommendation, the relationship data includes static relationship data, dynamic relationship data and case relationship data, and the step of determining relationship data between the plurality of management objects based on the multi-modal data further includes:
Extracting static relation data among management objects from the personnel static information data;
track fusion is carried out based on the dynamic track data to obtain fused dynamic track data, and dynamic relation data among management objects are extracted from the fused dynamic track data;
and extracting case relation data among management objects from the case file data.
In the above preferred technical solution of the ranking model building method for case cue recommendation, the step of generating the labels related to the plurality of management objects according to the multimodal data and the relationship data further includes:
and respectively generating a personal label of each management object and a relationship label among a plurality of management objects based on the personnel static information data, the fused dynamic track data and the knowledge graph.
In the above preferred technical solution of the ranking model establishment method for case cue recommendation, the step of generating a joint representation "based on the relationship data and the tag further includes:
generating a sequence structure vector with a set length by a random walk mode aiming at each management object in the knowledge graph, and recording the label of each walk object which is walked randomly, wherein each value in the sequence structure vector is the value of the relation label of the management object and each corresponding walk object;
Determining threat degree vectors by using a preset classification model based on the label of each passing object, wherein each value in the threat degree vectors is the threat degree of each passing object;
performing point multiplication on the sequence structure vector and the threat degree vector, inputting a result after the point multiplication into a dimension reduction algorithm, and obtaining a knowledge graph identifier after dimension reduction, thereby generating a joint representation of the knowledge graph and the label;
the classification model is used for representing the corresponding relation between the tag of the passing object and the threat degree.
In the above preferred technical solution of the ranking model establishing method for case cue recommendation, the step of establishing the ranking model "based on the joint representation" further includes:
and inputting the joint representation into a deep learning classification algorithm for training to obtain the ordering model.
In a second aspect of the present application, there is further provided a ranking model building apparatus for case cue recommendation, the apparatus including:
a data acquisition module for acquiring multi-modal data related to a plurality of management objects;
a relationship determination module for determining relationship data between a plurality of the management objects based on the multimodal data;
A tag generation module for generating tags related to the plurality of management objects according to the multimodal data and the relationship data;
a joint representation module for generating a joint representation based on the relationship data and the tag;
a model building module for building a ranking model for ranking the plurality of management objects based on the joint representation.
In the preferred technical solution of the ranking model establishing device for case cue recommendation, the relationship determining module is further configured to store the relationship data in a knowledge graph after determining the relationship data between the plurality of management objects.
In the above preferred technical solution of the ranking model building apparatus for case cue recommendation, the multimodal data includes: personnel static information data, dynamic track data and case file data.
In the preferred technical solution of the ranking model building apparatus for case cue recommendation, the relationship data includes static relationship data, dynamic relationship data and case relationship data, and the relationship determining module determines relationship data between a plurality of management objects based on the multimodal data by:
Extracting static relation data among management objects from the personnel static information data;
track fusion is carried out based on the dynamic track data to obtain fused dynamic track data, and dynamic relation data among management objects are extracted from the fused dynamic track data;
and extracting case relation data among management objects from the case file data.
In the preferred technical solution of the ranking model building apparatus for case cue recommendation, the tag generation module generates the tags related to the plurality of management objects according to the multimodal data and the relationship data by:
and respectively generating a personal label of each management object and a relationship label among a plurality of management objects based on the personnel static information data, the fused dynamic track data and the knowledge graph.
In the preferred technical solution of the ranking model building apparatus for case cue recommendation, the joint representation module generates a joint representation based on the relationship data and the tag by:
generating a sequence structure vector with a set length by a random walk mode aiming at each management object in the knowledge graph, and recording the label of each walk object which is walked randomly, wherein each value in the sequence structure vector is the value of the relation label of the management object and each corresponding walk object;
Determining threat degree vectors by using a preset classification model based on the label of each passing object, wherein each value in the threat degree vectors is the threat degree of each passing object;
performing point multiplication on the sequence structure vector and the threat degree vector, inputting a result after the point multiplication into a dimension reduction algorithm, and obtaining a knowledge graph identifier after dimension reduction, thereby generating a joint representation of the knowledge graph and the label;
the classification model is used for representing the corresponding relation between the tag of the passing object and the threat degree.
In the preferred technical solution of the ranking model establishing device for case cue recommendation, the model establishing module establishes the ranking model based on the joint representation by:
and inputting the joint representation into a deep learning classification algorithm for training to obtain the ordering model.
In a third aspect of the present application, there is also provided a case cue recommendation method, the method including:
extracting key factors aiming at an input case file and determining a plurality of management objects meeting the conditions based on the key factors;
ranking the plurality of management objects based on a ranking model;
Screening the plurality of sequenced management objects based on a business strategy;
outputting the screened management object.
In a fourth aspect of the present application, there is also provided a case cue recommendation apparatus, the apparatus including:
the recall module is used for extracting key factors from the input case files and determining a plurality of management objects meeting the conditions based on the key factors;
a ranking module for ranking the plurality of management objects based on a ranking model;
the screening module is used for screening the plurality of ordered management objects based on the business strategy;
and the output module is used for outputting the screened management objects.
In a fifth aspect of the present application, there is further provided a processing device, including a processor and a memory, where the memory is adapted to store a plurality of program codes, where the program codes are adapted to be loaded and executed by the processor to perform the arrangement model building method for case cue recommendation according to any one of the above preferred embodiments or the case cue recommendation method according to the above preferred embodiment.
In a sixth aspect of the present application, there is further provided a computer readable storage medium storing a plurality of program codes adapted to be loaded and executed by a processor to perform the arrangement model building method for case cue recommendation or the case cue recommendation method described in the above preferred embodiments.
The technical scheme provided by the invention has at least one or more of the following beneficial effects:
the multi-mode data of the management object is acquired to carry out learning modeling, so that the multi-mode mass data of the service party are fully utilized, and the description of the management object is more stereoscopic. When the built model is applied to a case clue recommendation system, compared with a recommendation system built by a single mode, more dimensional data can be utilized to give out a more stereoscopic recommendation result, so that the accuracy and coverage rate of recommendation can be greatly improved.
Furthermore, by adopting the personnel static information data, the dynamic track data and the case volume data as the multi-mode data, the management object can be depicted from multiple dimensions, so that the management object can be depicted more stereoscopically. For example, basic attributes of the management object can be described by personnel static information data, real behavior patterns of the management object can be described by dynamic track data, and the like.
Drawings
The method for establishing the sorting model, the method for recommending case clues, the device and the medium are described below with reference to the accompanying drawings. In the accompanying drawings:
FIG. 1 is a flow chart of a ranking model building method for case cue recommendation of the present invention;
FIG. 2 is a data flow diagram of a ranking model building process for case cue recommendation of the present invention;
FIG. 3 is a block diagram illustrating an exemplary embodiment of an ordering model set-up apparatus for case cue recommendation according to the present invention;
FIG. 4 is a flow chart of a case cue recommendation method of the present invention;
fig. 5 is a block diagram illustrating an embodiment of a case thread recommending apparatus according to the present invention.
List of reference numerals
11. A data acquisition module; 12. a relationship determination module; 13. a label generating module; 14. a joint representation module; 15. a model building module;
21. a recall module; 22. a sequencing module; 23. a screening module; 24. and an output module.
Detailed Description
Preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are merely for explaining the technical principles of the present invention, and are not intended to limit the scope of the present invention.
In the description of the present invention, a "module," "processor" may include hardware, software, or a combination of both. A module may comprise hardware circuitry, various suitable sensors, communication ports, memory, or software components, such as program code, or a combination of software and hardware. The processor may be a central processor, a microprocessor, an image processor, a digital signal processor, or any other suitable processor. The processor has data and/or signal processing functions. The processor may be implemented in software, hardware, or a combination of both. Non-transitory computer readable storage media include any suitable medium that can store program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, and the like. The term "a and/or B" means all possible combinations of a and B, such as a alone, B alone or a and B. The term "at least one A or B" or "at least one of A and B" has a meaning similar to "A and/or B" and may include A alone, B alone or A and B. The singular forms "a", "an" and "the" include plural referents.
Referring first to fig. 1 and 2, a description will be given of an arrangement model creation method for case cue recommendation of the present invention. FIG. 1 is a flowchart of a method for creating a ranking model for case cue recommendation according to the present invention; FIG. 2 is a data flow diagram of a ranking model building process for case cue recommendation of the present invention.
As shown in fig. 1 and fig. 2, in order to solve the problems of single logic and poor accuracy of the existing case cue recommendation system, the method for establishing the ranking model for case cue recommendation in the present application mainly includes:
s101, acquiring multi-mode data related to a plurality of management objects.
In one embodiment, the multimodal data includes: personnel static information data, dynamic track data and case file data. The personnel static information data refers to basic information of personnel, such as names, ages, sexes, residence places, native places and the like of management objects, and can be obtained from a system of business parties, such as a household registration system of a public security department. The dynamic track data refers to data capable of reflecting a space-time track of a management object, for example, a snapshot record (such as a face snapshot record, vehicle snapshot data, etc.), a track of a communication end, a transaction flow, a call record, a resident record, a train/plane trip record, etc. of the management object.
Wherein, the snapshot record of the management object can be determined by the data of the snapshot of cameras and video cameras installed at the positions of intersections, gates, elevators and the like. The communication terminal may be a mobile phone or other device with a communication function, which has a unique identifier, such as an IMSI (international mobile subscriber identity), a MAC address, and the communication terminal track may be determined by the sensed MAC address, IMSI. The position of the mobile phone can be determined through the MAC address, so that the track of the mobile phone is obtained. The MAC address may be acquired through a WIFI probe. The transaction flow can be called up through a bank, the record can be called up through an operator, the record of a living store can be called up through a hotel management system, the record of a train/plane trip can be called up through a traffic department, and the like.
Case file data refers to data that records the event process of the case, which data can also be obtained from the business party's system, such as from the file of the public security department.
Of course, the specific composition of the three data in the multi-mode data is not limited, and a person skilled in the art can adjust the specific ranges of the three data based on the specific application scenario, such as deleting the three data or replacing the three data with other related data based on the current data.
S103, based on the multi-mode data, determining relation data among the plurality of management objects, and storing the relation data in a knowledge graph.
In one possible implementation, taking still as an example that the multimodal data includes personnel static information data, dynamic track technical bureau and case volume data, the relationship data includes static relationship data, dynamic relationship data and case relationship data, step S103 may further include:
s1031, extracting static relation data established by management objects from personnel static information data; for example, from the name and residence information of the person, "Zhang san- & gtTongxiang- & gtLisi" relationship data is extracted. The extraction method is simple and will not be described in detail.
S1032, extracting case relation data among management objects from the case file data. For example, by existing natural language processing algorithms, including but not limited to: the LTAG algorithm, the CCG algorithm, the RST tree structure analysis algorithm, the reference resolution algorithm and the like analyze the language structure of the case description in the case file, extract the entity and the relation therefrom, for example, the relation of ' Zhang Sanzhang traffic hit- > Liqu ' is extracted from the case description of ' 15 points of 1 month and 10 days in 2020 ' in Zhang Jiang artificial intelligent island hit Liqu ', and the time (15 points of 1 month and 10 months in 2020) and the place (Zhang Jiang artificial intelligent island) of the case are taken as the attributes of the relation.
S1033, fusing the tracks to obtain fused dynamic track data aiming at the dynamic track data, and extracting dynamic relation data among management objects from the fused dynamic track data. Specifically, first, the related track data of the portrait and the vehicle based on snapshot is obtained from the portrait and the vehicle snapshot data through the existing algorithms such as a face detection algorithm, a face feature extraction algorithm, a feature clustering algorithm and the like. Taking a portrait as an example, the main data structure may include: face id, snapshot time, snapshot longitude, snapshot latitude, snapshot altitude, snapshot equipment address, etc. And secondly, fusing the related track data based on snapshot and other track data (such as transaction stream, call records, store records and train and plane trip records) in the dynamic track data through a graphic code association detection algorithm to obtain a fused dynamic track data track. The steps of the code association detection algorithm are briefly described as follows:
s1033-1, supplementing the historical space-time tracks of each management object by using a sequence supplementing method to form a supplemented historical space-time track, wherein the sequence supplementing method comprises but is not limited to generating an countermeasure network, tensor decomposition, linear interpolation, LSTM and the like.
S1033-2, converting the supplemented historical space-time track into a word vector by using a word vector model. Among them, word vector models include, but are not limited to, word2vec, VAE, WAE, auto-Encoders or SEQ-GAN, etc.
S1033-3, clustering word vectors by using a topic clustering model. Among them, topic cluster models include, but are not limited to, LDA, DBSCAN, etc.
S1033-4, constructing a subgraph of the management object corresponding to the word vector under each cluster and the historical space-time track thereof. The way of constructing the subgraph is as follows: and acquiring the pairwise combinations of the management objects, and calculating the space-time similarity of the historical space-time trajectories between the pairwise combinations as the weight of the edges. The method for calculating the space-time similarity includes, but is not limited to, euclidean distance, LCS distance and the like.
S1033-5, picking out the isolated point from the subgraph, wherein the definition of the isolated point is that the weights of all sides led out from the point are less than a set threshold value.
S1033-6, selecting representative nodes according to the centrality of the nodes for each subgraph.
S1033-7, calculating the space-time similarity of every two combinations for the representative node of each sub-graph, and fusing the sub-graphs with the space-time similarity exceeding a set threshold.
The process of fusing two space-time tracks is to use the same mark to identify two space-time tracks, for example, the mark 1 is used to identify a plurality of space-time tracks, for example, the mark 1 is used to identify the space-time track A, the space-time track B and the space-time track C, so that the space-time track A, the space-time track B and the space-time track C can be considered as tracks of the same management object.
While the foregoing description of the ccm algorithm is briefly described, those skilled in the art will appreciate that, in some other embodiments, some of the steps described above may be replaced or omitted, or those skilled in the art may use other ccm algorithms to perform fusion, for example, track fusion may be performed using a fusion method shown in publication No. CN111797295A or CN111695511 a.
And extracting the dynamic relation relationship among the management objects from the fused dynamic track data. The extraction modes of the dynamic relation data can be divided into the following three types: a. the relation of Zhang Sanzhuang, possession and mobile phone which is directly extracted by the graphic code association detection algorithm; b. the relationship discovery system based on the equivalence relationship automatically discovers relationships, such as: the relationship of Zhang Sanzhuang Lian Si is extracted from the store records; c. based on the relationship found by the traffic policy and statistics, for example, the traffic policy is: and counting call objects with the number of calls being more than 20 in a call record of a certain management object in the near week, so that the relation of Zhang Sanzhang frequent call and Liqu is found in the call record.
S1034, uniformly storing the relationship data between the management objects generated in the steps S1031-S1033 in the knowledge graph.
Compared with the conventional image code association detection algorithm in the prior art, the image code association detection algorithm can realize association of multiple types of management objects, such as association of multiple types of objects of human face-human body-vehicle-mobile phone-financial account numbers, by utilizing the knowledge graph and semantic topic clustering technology, so that the aim of describing and recommending management objects more stereoscopically and accurately is fulfilled.
S105, generating labels related to a plurality of management objects according to the multi-mode data and the knowledge graph.
In one possible implementation, the relationship label between the personal label of the management object and the management object may be calculated based on the business strategy from the personnel static information data, the fused dynamic track data and the knowledge graph by using the label factory system. For example, for a single management object, generating tags such as "elderly people", "high school" and the like from age, school data in personnel static information data; for another example, for a single management object, based on a business strategy that the number of times of occurrence at night is higher than 70% in the recent number of times of occurrence of a certain object, labels such as 'diurnal night out' are formed from the fused dynamic track data; for another example, for the relationship between a certain management object and other management objects, labels of 'frequent contact with the toxic person', 'primary relationship of important person (i.e. direct contact with important person)' and 'secondary relationship of important person (i.e. indirect contact with important person)' are formed from the knowledge graph; for another example, relationship labels such as "close relationship", "light relationship", "important relationship", etc. are generated for the relationships between different management objects, and different weights are given to the relationships between the management objects according to the relationship labels, for example, the weight of "important relationship" is 1, the weight of "close relationship" is 0.8, and the weight of "light relationship" is 0.3, etc. Among other things, label mill systems are common in the art, and the principles and processes for generating labels are not described in detail herein.
And S107, generating a joint representation based on the knowledge graph and the labels.
Because the final ordering model needs to input structured data, the labels in the obtained data belong to the structured data, the ordering model can be directly input, and the knowledge graph belongs to unstructured data, and the structured data needs to be converted. Thus, in one possible implementation, step S107 may further include: based on the knowledge graph and the relationship label between the management objects, the joint representation of the knowledge graph between the management objects and the relationship label between the management objects is learned through a graph embedding algorithm. The graph embedding algorithm specifically comprises the following steps:
s1071, generating sequence structure vectors with specified lengths by a random walk mode aiming at a certain node, namely a certain management object, in the knowledge graph, and sequentially recording the labels (hereinafter or simply referred to as ids) of each passed management object (namely passed objects). Each value in the sequence structure vector is a label value of each edge of the random walk of the node (namely, a relation label value of the management object and each corresponding passing object). The relationship label is given a value, that is, the weight, for example, the "relationship requiring important attention" is labeled as 1, the "close relationship" is labeled as 0.8, the "light relationship" is labeled as 0.3, and the like according to the weight. If the random walks 5 times, the sequence structure vector may be generated, for example: (1,0.3,0.8,0.3,0.3), the id of each pass object recorded is: (id 1, id2, id3, id2, id 4).
S1072, based on labels carried by each passing object, threat degrees of each passing object are respectively determined by using a preset classification model, and threat degree vectors are obtained. The classification model is used for representing the corresponding relation between the tags of the passing objects and the threat degree, in a preferred implementation manner, the classification model can be built according to various tags on the management object in the history file data based on the history file data of the criminal investigation party, wherein the classification model can be an intelligent grading card model, and the specific building method of the intelligent grading card model can refer to patent application with publication number of CN111563810A or CN 111898675A. The threat level is a quantitative indicator, such as a value between 0 and 1, which can be used to indicate the suspicious level of passing objects. For example, the threat degree vector composed of threat degrees of the random walk object is: (0,0.3,0.7,0.3,1)
S1073, carrying out dot multiplication on the sequence structure vector and the threat degree vector, and inputting the dot multiplication result into a dimension reduction algorithm to obtain the dimension reduced knowledge graph identification. For example, the two vector points are multiplied by (1,0.3,0.8,0.3,0.3) · (0,0.3,0.7,0.3,1) = (0,0.09,0.56,0.09,0.3), the result after the point multiplication is input into the PCA automatic dimension reduction algorithm, and the dimension-reduced knowledge graph identification is obtained, and at this time, the knowledge graph is converted into structured information, and can be jointly represented with the label.
Of course, in addition to the graph embedding algorithm described above, one skilled in the art may use other graph embedding algorithms to replace the algorithm described above, or to replace certain steps of the graph embedding algorithm described above, to adjust the interpretability of the ranking model. For example, the Node2Vec algorithm is used to replace the graph embedding algorithm of the application, or a logistic regression model and a DNN model are used to replace the classification model, or an automatic encoder, a VAE and other algorithms are used to replace the PCA dimension reduction algorithm. However, compared with the algorithm provided by the application, the alternative algorithm may have the defects of poor interpretability, incapability of combining with an actual security scene and the like.
Generally criminal investigation often requires that the recommended results for the clues be interpretable and traceable. In the implementation process of the graph embedding algorithm, the algorithm is configurable and the features can be processed automatically in the model building process by combining a random walk, a classification model and a dimension reduction algorithm based on the data features and the business features of the criminal investigation industry. By further adopting the technical scheme of combining the random walk, the intelligent scoring card model and the PCA dimension-reducing algorithm, the random walk, the intelligent scoring card model and the PCA dimension-reducing algorithm have interpretability in the application process, so that the whole graph embedding algorithm has interpretability, and the development and training processes of the algorithm are highly combined with criminal investigation data, so that the algorithm model is more close to the actual use scene of criminal investigation while the high interpretability is maintained, and the accuracy equivalent to or higher than that of the existing model is realized in the criminal investigation field.
S109, establishing a sequencing model based on the joint representation.
In one possible embodiment, the step S109 further includes: the joint representation is input into a deep learning classification algorithm for training to obtain a ranking model for ranking a plurality of management objects. The process of training the joint representation input into the deep learning classification algorithm to obtain the ranking model is more conventional, and will not be described in detail herein.
Of course, the person skilled in the art may replace the modeling algorithm described above, so as to obtain a ranking model that is more suitable for a specific application scenario.
In summary, according to the method for establishing the ordering model for case clue recommendation, for data of different modes, the data are changed into the relational data and the labels through operations such as track fusion, relational calculation and label calculation, the relational data and the labels are respectively stored in a knowledge graph and a structured database, the joint representation of the knowledge graph and the labels is obtained through graph embedding algorithm learning, and the ordering model is established based on the joint representation. The interpretability and accuracy of the ranking model can be adjusted in the process by changing the choice of graph embedding algorithm and modeling algorithm.
When the sorting model is applied to clue recommendation, compared with a single-mode recommendation system, the sorting model can utilize more dimensionality data, and recommendation results are more stereoscopic. Such as: the personnel static information data can be used for describing the basic attribute of the management object, and the dynamic track data can be used for describing the real behavior mode of the management object, so that the recommendation accuracy and coverage rate can be improved. Through a mode of combining multi-mode learning and a service strategy of a criminal investigation party, a perception and cognition decision is made, the recommendation is more numerous, and the recommendation effect is better.
It should be noted that, although the steps are described in the above-mentioned sequential order in the above-mentioned embodiments, it will be understood by those skilled in the art that, in order to achieve the effects of the present embodiment, the steps need not be performed in such an order, and may be performed simultaneously (in parallel) or in reverse order, and these simple variations are all within the scope of the present invention. For example, steps S1031 to S1033 may be performed simultaneously, or may be performed in any order.
In an alternative embodiment, although the above embodiment is described in connection with storing a plurality of determined relationship data in a knowledge-graph, this embodiment is not a complete one, and one skilled in the art can choose whether to store the relationship data in the knowledge-graph based on a specific application scenario. For example, when the data amount is not large, the determined relationship data may not be stored in the knowledge graph, but directly participate in the calculation of the tag and the generation of the joint representation in the manner of a relationship table. Of course, the relation type data are rich, and the criminal investigation industry data are different from other industry data, and if the knowledge graph is not used, the calculation speed of the model can be influenced.
Referring to fig. 3, a ranking model establishing apparatus for case cue recommendation of the present application will be described. FIG. 3 is a block diagram illustrating an embodiment of an apparatus for creating a ranking model for case cue recommendation according to the present invention.
As shown in fig. 3, the ranking model building device for case cue recommendation in the present application mainly includes: a data acquisition module 11 for acquiring multi-modal data related to a plurality of management objects; a relationship determination module 12 for determining relationship data between a plurality of management objects based on the multimodal data, and storing the relationship data in a knowledge graph; a tag generation module 13 for generating tags related to a plurality of management objects according to the multimodal data and the knowledge graph; a joint representation module 14 for generating a joint representation based on the knowledge-graph and the labels; a model building module 15 for building a ranking model based on the joint representation. In one embodiment, the description of the specific implementation functions may be referred to in steps S101-S109.
In one embodiment, the multimodal data includes: personnel static information data, dynamic track data and case file data.
In one embodiment, the relationship data includes static relationship data, dynamic relationship data, and case relationship data, and the relationship determination module 12 determines relationship data between the plurality of management objects based on the multimodal data by: extracting static relation data among management objects from personnel static information data; track fusion is carried out based on the dynamic track data to obtain fused dynamic track data, and dynamic relation data among management objects are extracted from the fused dynamic track data; case relation data among management objects is extracted from the case file data. The description of the specific implementation function can be found in the above step S103.
In one embodiment, the tag generation module 13 generates tags related to a plurality of management objects based on the multimodal data and the knowledge graph by: based on the personnel static information data, the fused dynamic track data and the knowledge graph, respectively generating a personal label of each management object and a relationship label among a plurality of management objects. The description of the specific implementation function can be found in the above step S105.
In one embodiment, the joint representation module 14 generates the joint representation based on the knowledge-graph and the labels by: generating a sequence structure vector with a set length by a random walk mode aiming at each management object in the knowledge graph, and recording the label of each walk object which is walked randomly, wherein each value in the sequence structure vector is the value of the relation label of the management object and each corresponding walk object; determining threat degree vectors by using a preset classification model based on the label of each passing object, wherein each value in the threat degree vectors is the threat degree of each passing object; performing point multiplication on the sequence structure vector and the threat degree vector, inputting the result after the point multiplication into a dimension reduction algorithm, and obtaining a dimension-reduced knowledge graph identifier, thereby generating a joint representation of the knowledge graph and the label; the classification model is used for representing the corresponding relation between the tags of the passing objects and the threat degree. The description of the specific implementation function can be found in the above step S107.
In one embodiment, the model building module 15 builds the ranking model based on the joint representation by: and inputting the joint representation into a deep learning classification algorithm for training to obtain a sorting model. The description of the specific implementation function can be found in the above step S109.
The technical principles of the above embodiments of the ranking model establishing device for case thread recommendation and the technical problems to be solved and the technical effects to be produced are similar, and those skilled in the art can clearly understand that, for convenience and brevity of description, the specific working process and related description of the ranking model establishing device for case thread recommendation may refer to the description of the embodiments of the ranking model establishing method for case thread recommendation, and will not be repeated herein.
Referring to fig. 4, a case thread recommendation method of the present application will be described. Fig. 4 is a flowchart of a case cue recommendation method according to the present invention.
As shown in fig. 4, the case cue recommendation method of the present application mainly includes the following steps:
s201, extracting key factors aiming at the input case file and determining a plurality of management objects meeting the conditions based on the key factors.
In one possible implementation, key factors in the input case file are extracted through a grammar tree model, and management objects which possibly meet the conditions are primarily screened for the key factors. Key factors include, but are not limited to: the identity of the suspect, alarm receiving record, time and place in the record, etc.
S203, sorting the plurality of management objects based on the sorting model.
In one possible implementation manner, the management objects of the primary screening are input into the sorting model obtained by the sorting model building method for case clue recommendation, and the coincidence degree of each management object is output through the sorting model and sorted according to the coincidence degree. Of course, before using the ranking model, training of the ranking model is required. The training mode may be to train the model by taking the case data already set as the input and output of the model.
S205, screening the sorted multiple management objects based on the service policy.
In one possible implementation, after the possible management objects are ordered, not all the management objects are required by the service party, and at this time, the service policy provided by the service party needs to be used to further screen the management objects that are screened out, so as to reduce the scope of the management objects. The service policy is a screening condition provided by the service party, such as screening management objects in daytime and nighttime, or screening management objects with frequent calls.
S207, outputting the screened management objects.
In one possible implementation manner, after the management objects are further screened, the screened management object list can be output according to the form of the coincidence ordering, so that the recommendation of the case clues is realized.
According to the case clue recommending method, compared with a single-mode recommending system, the case clue recommending method can utilize more dimensionality data, and recommending results are more stereoscopic, so that recommending accuracy and coverage rate can be improved. The ordered management objects are screened based on the service strategies, so that the multi-mode data can be combined with the service strategies of criminals, perception and cognition decisions are communicated, the recommended is more, and the recommended effect is better.
Next, referring to fig. 5, a case thread recommending apparatus of the present application will be described. Fig. 5 is a block diagram illustrating an embodiment of a case thread recommendation apparatus according to the present invention.
As shown in fig. 5, the case cue recommendation device of the present application mainly includes: a recall module 21 for extracting key factors for the input case volumes and determining a plurality of management objects conforming to the conditions based on the key factors; a ranking module 22 for ranking the plurality of management objects based on the ranking model; a screening module 23, configured to screen the sorted multiple management objects based on a service policy; and an output module 24 for outputting the screened management object. In one embodiment, the description of the specific implementation functions may refer to step S201-step S205.
The above case thread recommending apparatus is used for executing the case thread recommending method embodiment, and the technical principles of the two, the technical problems to be solved and the technical effects to be produced are similar, and those skilled in the art can clearly understand that, for convenience and brevity of description, the specific working process and the related description of the case thread recommending apparatus can refer to the description of the case thread recommending method embodiment, and the description is not repeated here.
It will be appreciated by those skilled in the art that the present invention may implement all or part of the above-described methods in one embodiment, or may be implemented by a computer program for instructing the relevant hardware, and the computer program may be stored in a computer readable storage medium, where the computer program when executed by a processor implements the steps of the respective method embodiments described above. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, executable files or in some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, a medium, a USB flash disk, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier wave signal, a telecommunication signal, a software distribution medium, and so forth. It should be noted that the content of the computer readable medium can be appropriately increased or decreased according to the requirements of the jurisdiction's jurisdiction and the patent practice, for example, in some jurisdictions, the computer readable medium does not include electrical carrier signals and telecommunication signals according to the jurisdiction and the patent practice.
Various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functions of some or all of the components in a server, client, according to embodiments of the invention, may be implemented in practice using a microprocessor or Digital Signal Processor (DSP). The present invention may also be embodied as a device or apparatus program (e.g., a PC program and a PC program product) for performing part or all of the methods described herein. Such a program embodying the present invention may be stored on a PC readable medium, or may have the form of one or more signals. Such signals may be downloaded from an internet website, provided on a carrier signal, or provided in any other form.
Further, the invention also provides a computer readable storage medium. In one embodiment of the computer-readable storage medium according to the present invention, the computer-readable storage medium may be configured to store a program for performing the arrangement model creation method for case cue recommendation and/or the case cue recommendation method of the above-described method embodiment, the program being loadable and executable by a processor to implement the arrangement model creation method for case cue recommendation and/or the case cue recommendation method described above. For convenience of explanation, only those portions of the embodiments of the present invention that are relevant to the embodiments of the present invention are shown, and specific technical details are not disclosed, please refer to the method portions of the embodiments of the present invention. The computer readable storage medium may be a storage device including various electronic devices, and optionally, the computer readable storage medium in the embodiments of the present invention is a non-transitory computer readable storage medium.
Further, the invention also provides a processing device. In an embodiment of the processing device according to the present invention, the processing device includes a processor and a memory, the memory may be configured to store a program for executing the arrangement model creation method for case cue recommendation and/or the case cue recommendation method of the above-described method embodiment, and the processor may be configured to execute the program in the memory, including, but not limited to, the program for executing the arrangement model creation method for case cue recommendation and/or the case cue recommendation method of the above-described method embodiment. For convenience of explanation, only those portions of the embodiments of the present invention that are relevant to the embodiments of the present invention are shown, and specific technical details are not disclosed, please refer to the method portions of the embodiments of the present invention. The processing means may be an apparatus device formed including various electronic devices.
It should be understood that, since the respective modules are merely set for illustrating the functional units of the apparatus of the present invention, the physical devices corresponding to the modules may be the processor itself, or a part of software in the processor, a part of hardware, or a part of a combination of software and hardware. Accordingly, the number of individual modules in the figures is merely illustrative.
Those skilled in the art will appreciate that the various modules in the apparatus may be adaptively split or combined. Such splitting or combining of specific modules does not cause the technical solution to deviate from the principle of the present invention, and therefore, the technical solution after splitting or combining falls within the protection scope of the present invention.
Thus far, the technical solution of the present invention has been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of protection of the present invention is not limited to these specific embodiments. Equivalent modifications and substitutions for related technical features may be made by those skilled in the art without departing from the principles of the present invention, and such modifications and substitutions will fall within the scope of the present invention.
Claims (14)
1. A method for creating a ranking model for case cue recommendation, the method comprising:
acquiring multi-modal data related to a plurality of management objects;
determining relationship data between a plurality of the management objects based on the multimodal data;
generating labels related to a plurality of management objects according to the multi-modal data and the relation data;
Generating a joint representation based on the relationship data and the tag;
based on the joint representation, establishing a ranking model for ranking a plurality of the management objects;
after the step of "determining relationship data between a plurality of the management objects", the method further includes:
storing the relation data in a knowledge graph;
the step of generating a joint representation based on the relationship data and the tag further comprises:
generating a sequence structure vector with a set length by a random walk mode aiming at each management object in the knowledge graph, and recording the label of each walk object which is walked randomly, wherein each value in the sequence structure vector is the value of the relation label of the management object and each corresponding walk object;
determining threat degree vectors by using a preset classification model based on the label of each passing object, wherein each value in the threat degree vectors is the threat degree of each passing object;
performing point multiplication on the sequence structure vector and the threat degree vector, inputting a result after the point multiplication into a dimension reduction algorithm, and obtaining a knowledge graph identifier after dimension reduction, thereby generating a joint representation of the knowledge graph and the label;
The classification model is used for representing the corresponding relation between the tag of the passing object and the threat degree.
2. The ranking model building method for case cue recommendation according to claim 1, wherein the multi-modal data comprises: personnel static information data, dynamic track data and case file data.
3. The method for creating a ranking model for case cue recommendation according to claim 2, wherein the relationship data includes static relationship data, dynamic relationship data and case relationship data, and the step of determining relationship data between a plurality of the management objects based on the multi-modal data further comprises:
extracting static relation data among management objects from the personnel static information data;
track fusion is carried out based on the dynamic track data to obtain fused dynamic track data, and dynamic relation data among management objects are extracted from the fused dynamic track data;
and extracting case relation data among management objects from the case file data.
4. The ranking model establishing method for case cue recommendation according to claim 3, wherein the step of generating tags related to the plurality of management objects based on the multimodal data and the relationship data further comprises:
And respectively generating a personal label of each management object and a relationship label among a plurality of management objects based on the personnel static information data, the fused dynamic track data and the knowledge graph.
5. The ranking model establishing method for case cue recommendation according to claim 1, wherein the step of establishing a ranking model based on the joint representation further comprises:
and inputting the joint representation into a deep learning classification algorithm for training to obtain the ordering model.
6. An ordering model building device for case clue recommendation, characterized in that the device comprises:
a data acquisition module for acquiring multi-modal data related to a plurality of management objects;
a relationship determination module for determining relationship data between a plurality of the management objects based on the multimodal data;
a tag generation module for generating tags related to the plurality of management objects according to the multimodal data and the relationship data;
a joint representation module for generating a joint representation based on the relationship data and the tag;
a model building module for building a ranking model for ranking a plurality of the management objects based on the joint representation;
The relationship determination module is further used for storing the relationship data in a knowledge graph after determining the relationship data among a plurality of management objects;
the joint representation module generates a joint representation based on the relationship data and the tag by:
generating a sequence structure vector with a set length by a random walk mode aiming at each management object in the knowledge graph, and recording the label of each walk object which is walked randomly, wherein each value in the sequence structure vector is the value of the relation label of the management object and each corresponding walk object;
determining threat degree vectors by using a preset classification model based on the label of each passing object, wherein each value in the threat degree vectors is the threat degree of each passing object;
performing point multiplication on the sequence structure vector and the threat degree vector, inputting a result after the point multiplication into a dimension reduction algorithm, and obtaining a knowledge graph identifier after dimension reduction, thereby generating a joint representation of the knowledge graph and the label;
the classification model is used for representing the corresponding relation between the tag of the passing object and the threat degree.
7. The ranking model establishing apparatus for case cue recommendation of claim 6 wherein the multimodal data comprises: personnel static information data, dynamic track data and case file data.
8. The ranking model establishing apparatus for case cue recommendation according to claim 7, wherein the relationship data includes static relationship data, dynamic relationship data, and case relationship data, and the relationship determination module determines relationship data between a plurality of the management objects based on the multimodal data by:
extracting static relation data among management objects from the personnel static information data;
track fusion is carried out based on the dynamic track data to obtain fused dynamic track data, and dynamic relation data among management objects are extracted from the fused dynamic track data;
and extracting case relation data among management objects from the case file data.
9. The ranking model building apparatus for case cue recommendation according to claim 8, wherein the tag generation module generates tags related to the plurality of management objects from the multimodal data and the relationship data by:
And respectively generating a personal label of each management object and a relationship label among a plurality of management objects based on the personnel static information data, the fused dynamic track data and the knowledge graph.
10. The ranking model establishing apparatus for case cue recommendation of claim 6 wherein the model establishing module establishes a ranking model based on the joint representation by:
and inputting the joint representation into a deep learning classification algorithm for training to obtain the ordering model.
11. A case cue recommendation method, the method comprising:
extracting key factors aiming at an input case file and determining a plurality of management objects meeting the conditions based on the key factors;
ranking the plurality of management objects based on a ranking model;
screening the plurality of sequenced management objects based on a business strategy;
outputting the screened management objects;
wherein the ranking model is built based on the ranking model building method for case cue recommendation of any one of claims 1 to 5.
12. A case thread recommendation device, the device comprising:
The recall module is used for extracting key factors from the input case files and determining a plurality of management objects meeting the conditions based on the key factors;
a ranking module for ranking the plurality of management objects based on a ranking model;
the screening module is used for screening the plurality of ordered management objects based on the business strategy;
the output module is used for outputting the screened management objects;
wherein the ranking model is built based on the ranking model building method for case cue recommendation of any one of claims 1 to 5.
13. A processing device comprising a processor and a memory, the memory being adapted to store a plurality of program codes, characterized in that the program codes are adapted to be loaded and executed by the processor to perform the ranking model establishing method for case thread recommendation of any one of claims 1 to 5 or the case thread recommendation method of claim 11.
14. A computer readable storage medium storing a plurality of program codes, wherein the program codes are adapted to be loaded and executed by a processor to perform the ranking model establishing method for case thread recommendation of any one of claims 1 to 5 or the case thread recommendation method of claim 11.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110181900.0A CN112925899B (en) | 2021-02-09 | 2021-02-09 | Ordering model establishment method, case clue recommendation method, device and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110181900.0A CN112925899B (en) | 2021-02-09 | 2021-02-09 | Ordering model establishment method, case clue recommendation method, device and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112925899A CN112925899A (en) | 2021-06-08 |
| CN112925899B true CN112925899B (en) | 2023-06-23 |
Family
ID=76171483
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110181900.0A Active CN112925899B (en) | 2021-02-09 | 2021-02-09 | Ordering model establishment method, case clue recommendation method, device and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112925899B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113961571B (en) * | 2021-12-22 | 2022-03-22 | 太极计算机股份有限公司 | Multi-mode data sensing method and device based on data probe |
| CN114090909A (en) * | 2022-01-18 | 2022-02-25 | 深圳前海中电慧安科技有限公司 | Graph code joint detection correlation method and device, computer equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109684470A (en) * | 2019-01-09 | 2019-04-26 | 中国科学技术大学 | Legal information recommended method and device, storage medium and electronic equipment |
| CN110008986A (en) * | 2019-02-19 | 2019-07-12 | 阿里巴巴集团控股有限公司 | The recognition methods of batch risk case, device and electronic equipment |
| CN110378126A (en) * | 2019-07-26 | 2019-10-25 | 北京中科微澜科技有限公司 | A kind of leak detection method and system |
| CN111241241A (en) * | 2020-01-08 | 2020-06-05 | 平安科技（深圳）有限公司 | Case retrieval method, device and equipment based on knowledge graph and storage medium |
| CN112333706A (en) * | 2019-07-16 | 2021-02-05 | 中国移动通信集团浙江有限公司 | Internet of things equipment anomaly detection method and device, computing equipment and storage medium |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104346355B (en) * | 2013-07-26 | 2019-03-08 | 南京中兴力维软件有限公司 | The intelligent search method and its system of serial public security case |
| US10121370B2 (en) * | 2014-09-20 | 2018-11-06 | Mohamed Roshdy Elsheemy | Comprehensive traffic control system |
| CN104572615A (en) * | 2014-12-19 | 2015-04-29 | 深圳中创华安科技有限公司 | Method and system for on-line case investigation processing |
| CN106126680A (en) * | 2016-06-29 | 2016-11-16 | 北京互信互通信息技术有限公司 | A kind of video image reconnaissance method and system |
| CN111209776A (en) * | 2018-11-21 | 2020-05-29 | 杭州海康威视系统技术有限公司 | Method, device, processing server, storage medium and system for identifying pedestrians |
| CN111090779A (en) * | 2019-03-01 | 2020-05-01 | 王文梅 | Cloud storage and retrieval analysis method for case-handling exploration evidence-taking data |
| CN111143602B (en) * | 2019-12-24 | 2023-05-02 | 云粒智慧科技有限公司 | Case clue association method, system, electronic equipment and storage medium |
| CN111401775A (en) * | 2020-03-27 | 2020-07-10 | 深圳壹账通智能科技有限公司 | Information analysis method, device, equipment and storage medium of complex relation network |
| CN111666495B (en) * | 2020-06-05 | 2023-08-11 | 北京百度网讯科技有限公司 | Case recommending method, device, equipment and storage medium |
| CN112101234B (en) * | 2020-09-16 | 2022-11-22 | 上海寰创通信科技股份有限公司 | Detection code matching processing method and graph code joint detection system |
-
2021
- 2021-02-09 CN CN202110181900.0A patent/CN112925899B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109684470A (en) * | 2019-01-09 | 2019-04-26 | 中国科学技术大学 | Legal information recommended method and device, storage medium and electronic equipment |
| CN110008986A (en) * | 2019-02-19 | 2019-07-12 | 阿里巴巴集团控股有限公司 | The recognition methods of batch risk case, device and electronic equipment |
| CN112333706A (en) * | 2019-07-16 | 2021-02-05 | 中国移动通信集团浙江有限公司 | Internet of things equipment anomaly detection method and device, computing equipment and storage medium |
| CN110378126A (en) * | 2019-07-26 | 2019-10-25 | 北京中科微澜科技有限公司 | A kind of leak detection method and system |
| CN111241241A (en) * | 2020-01-08 | 2020-06-05 | 平安科技（深圳）有限公司 | Case retrieval method, device and equipment based on knowledge graph and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112925899A (en) | 2021-06-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110147722A (en) | A kind of method for processing video frequency, video process apparatus and terminal device | |
| Tracewski et al. | Repurposing a deep learning network to filter and classify volunteered photographs for land cover and land use characterization | |
| CN104915351A (en) | Picture sorting method and terminal | |
| CN112925899B (en) | Ordering model establishment method, case clue recommendation method, device and medium | |
| CN110232104B (en) | Data display method and device, storage medium and computer equipment | |
| CN109902681B (en) | User group relation determining method, device, equipment and storage medium | |
| US9665773B2 (en) | Searching for events by attendants | |
| US20240037142A1 (en) | Systems and methods for filtering of computer vision generated tags using natural language processing | |
| CN111383004A (en) | Method for extracting entity position of digital currency, method for extracting information and device thereof | |
| CN105518644A (en) | Method for processing and displaying real-time social data on map | |
| CN111695019A (en) | Method and device for identifying associated account | |
| CN114219971A (en) | Data processing method, data processing equipment and computer readable storage medium | |
| CN113268649A (en) | Thread monitoring method and system based on diversified data fusion | |
| CN112766119A (en) | Method for accurately identifying strangers and constructing community security based on multi-dimensional face analysis | |
| CN114331473A (en) | Method and device for identifying telecommunication fraud event and computer-readable storage medium | |
| CN105930430B (en) | Real-time fraud detection method and device based on non-accumulative attribute | |
| WO2021114615A1 (en) | Method, apparatus, and device for visualization of behavior risk identification, and storage medium | |
| CN112306985A (en) | Digital retina multi-modal feature combined accurate retrieval method | |
| CN116028467A (en) | Intelligent service big data modeling method, system, storage medium and computer equipment | |
| CN114386415A (en) | Method for constructing digital-image fusion library and related equipment | |
| CN114332599A (en) | Image recognition method, image recognition device, computer equipment, storage medium and product | |
| Yang et al. | TAPESTRY: A blockchain based service for trusted interaction online | |
| CN112416922A (en) | Group partner association data mining method, device, equipment and storage medium | |
| CN112333182A (en) | File processing method, device, server and storage medium | |
| CN112364066B (en) | Communication method and system based on shared information |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |