Context-based object matching method and apparatus, terminal device, and storage medium

Abstract

The application relates to the technical field of object matching, in particular to a context-based object matching method and device, terminal equipment and a storage medium, wherein the method comprises the following steps: based on context information of a target object in an information source, at least one attribute of the target object is extracted, and an attribute set of the target object is constructed according to the at least one attribute of the target object; then, attribute sets of at least one candidate object in a preset database are acquired, similarity matching is performed on the attribute set of the target object and the attribute sets of the candidate objects, and the similarity of the target object and the at least one candidate object is obtained; and finally, a matching result between the target object and the at least one candidate object is determined. Therefore, the problems of matching limitations, high error rate and incomplete information completion in processing object recognition and alignment in multi-source heterogeneous data due to the fact that the prior art relies on fixed attributes, context utilization is insufficient and correlation reasoning capability is insufficient in the related art are solved.

Description

Technical Field

[0001] This application relates to the field of object matching technology, and in particular to a context-based object matching method, apparatus, terminal device and storage medium. Background Technology

[0002] In the field of data integration and object recognition, the same object in different data sources often has inconsistent names due to aliases and abbreviations, or such ambiguous issues as "Zhang San" corresponding to multiple different individuals, making it difficult to accurately identify the object's identity.

[0003] In related technologies, traditional methods rely on fixed attributes such as name and basic identity information for matching, or match only by name or a single attribute.

[0004] However, in related technologies, matching relies on fixed attributes such as name and basic identity information. When the direct contextual information provided in the text, such as organization, project, or research field, is insufficient, it is difficult to infer the identity of the object through indirect association, leading to missing information or misjudgment. Matching based solely on name or a single attribute cannot handle homonyms or differences in attribute descriptions, resulting in low matching accuracy. Finally, the ability of association reasoning is limited. For objects with missing contextual information, existing methods cannot utilize knowledge graphs or associations such as collaborative projects or social relationships for chain inference, and cannot indirectly supplement the missing information of the target object through the attributes of known objects, resulting in low information completion efficiency and narrow coverage.

[0005] In summary, existing technologies rely on fixed attributes, fail to fully utilize context, and lack sufficient associative reasoning capabilities, resulting in limitations in matching, high misjudgment rates, and incomplete information completion when processing object recognition and alignment in multi-source heterogeneous data; these issues urgently need improvement. Summary of the Invention

[0006] This application provides a context-based object matching method, apparatus, terminal device, and storage medium to solve the problems in related technologies, such as limited matching, high misjudgment rate, and incomplete information completion, which are caused by the reliance on fixed attributes, insufficient utilization of context, and insufficient associative reasoning ability in the prior art when processing object recognition and alignment in multi-source heterogeneous data.

[0007] The first aspect of this application provides a context-based object matching method, comprising the following steps: extracting at least one attribute of the target object based on context information of the target object in an information source, and constructing an attribute set of the target object based on the at least one attribute of the target object; obtaining at least one candidate object from a preset database; obtaining the attribute set of the at least one candidate object in the preset database, performing similarity matching between the attribute set of the target object and the attribute set of the candidate object to obtain the similarity between the target object and the at least one candidate object; and determining the matching result between the target object and the at least one candidate object based on the similarity.

[0008] Through the above technical solution, the embodiments of this application can obtain the attribute sets of the target object and the candidate object respectively, and then perform similarity matching on the attribute set of the target object and the attribute set of the candidate object to obtain the similarity between the target object and the candidate object, thereby determining whether the target object and the candidate object match. This realizes the transformation of implicit relationships between objects into computable attribute dimensions through dynamic integration of multi-source data, effectively avoiding matching limitations caused by name ambiguity and missing attributes, and significantly improving the accuracy and robustness of object matching in complex scenarios.

[0009] Optionally, in one embodiment of this application, obtaining candidate objects from a preset database includes: obtaining the name of the target object; and using the name as an index, querying the preset database for at least one candidate object that satisfies a preset name condition.

[0010] Through the above technical solution, the embodiments of this application can filter out several objects with completely identical names or those that are determined to be similar by the algorithm as candidate objects in the database, thereby narrowing the scope for subsequent similarity matching.

[0011] Optionally, in one embodiment of this application, the step of performing similarity matching between the attribute set of the target object and the attribute set of the candidate objects to obtain the similarity between the target object and at least one candidate object includes: performing semantic vectorization processing on the attribute set of the target object and the attribute set of the at least one candidate object based on a preset language model to obtain the attribute vector set of the target object and the attribute vector set of the at least one candidate object respectively; comparing the attribute vector set of the target object with the attribute vector set of the at least one candidate object, and calculating the similarity between the target object and the at least one candidate object based on the comparison result.

[0012] Through the above technical solution, the embodiments of this application can first perform quantization processing of the attribute set based on the large language model, and then compare the quantified attribute set to realize the similarity calculation of the target and the candidate object, effectively reducing the problem of deviation in similarity calculation due to the inability of attributes to correspond.

[0013] Optionally, in one embodiment of this application, after constructing the attribute set of the target object, the method further includes: determining a new target object from multiple publicly available target information sources; extracting the attributes of the new target object based on the context information of the new target object; and updating the attribute set using the attributes of the new target object.

[0014] Through the above technical solution, the embodiments of this application can determine the target object from several publicly available information sources, then extract the attributes of the target object, and update the attribute set of the target object, thereby making the attribute set of the target object more complete and increasing the matching accuracy between the target object and the candidate objects in the database.

[0015] Optionally, in one embodiment of this application, after constructing the attribute set of the target object, the method further includes: obtaining the association relationship of the target object, and obtaining at least one associated object based on the association relationship; obtaining the attributes of the at least one associated object, and constructing the attribute set of the at least one associated object based on the attributes of the at least one associated object; and updating the attribute set of the target object based on the attribute set of the at least one associated object.

[0016] Through the above technical solution, the embodiments of this application can update the attribute set of the target object based on the attribute set of the associated object, further improve the attribute set, make it contain more valuable information, thereby improving the accuracy and comprehensiveness of subsequent matching with candidate objects in the database.

[0017] Optionally, in one embodiment of this application, obtaining the association relationship of the target object and obtaining at least one associated object based on the association relationship includes: obtaining a graph structure containing the target object, the graph structure being constructed based on the context information of the target object and used to describe the association relationship between the target object and the at least one associated object; and determining the at least one associated object based on the graph structure.

[0018] Through the above technical solution, the embodiments of this application can obtain related objects based on the obtained graph structure containing the target object, thereby realizing the mining of multi-degree relationships and the revelation of implicit relationships; the technology of obtaining related objects through graph structure is significantly better than traditional association analysis methods in multiple dimensions.

[0019] Optionally, in one embodiment of this application, after constructing the attribute set of the at least one associated object, the method further includes: obtaining at least one secondary associated object for each of the at least one associated object; constructing the attribute set of the at least one secondary associated object; and updating the attribute set of the target object based on the attribute set of the at least one secondary associated object.

[0020] Through the above technical solutions, the embodiments of this application can construct a composite attribute set during the object matching process, organically integrating the object's own attributes and relationships. Unlike traditional methods that rely solely on a fixed single attribute for matching, this approach achieves dynamic integration of multi-source data, transforming implicit relationships between objects into computable attribute dimensions. This effectively solves the matching limitations caused by name ambiguity and missing attributes. By constructing a composite feature space that includes semantic and structural relationships, the accuracy and robustness of object matching in complex scenarios are significantly improved.

[0021] A second aspect of this application provides a context-based object matching apparatus, comprising: a target object attribute set construction module, configured to extract at least one attribute of the target object based on context information of the target object in an information source, and construct an attribute set of the target object based on the at least one attribute of the target object; a candidate object acquisition module, configured to acquire at least one candidate object from a preset database; a similarity calculation module, configured to acquire the attribute set of the at least one candidate object in the preset database, perform similarity matching between the attribute set of the target object and the attribute set of the candidate object, and obtain the similarity between the target object and the at least one candidate object; and a matching module, configured to determine the matching result between the target object and the at least one candidate object based on the similarity.

[0022] Through the above technical solution, the embodiments of this application can obtain the attribute sets of the target object and the candidate object respectively, and then perform similarity matching on the attribute set of the target object and the attribute set of the candidate object to obtain the similarity between the target object and the candidate object, thereby determining whether the target object and the candidate object match. This realizes the transformation of implicit relationships between objects into computable attribute dimensions through dynamic integration of multi-source data, effectively avoiding matching limitations caused by name ambiguity and missing attributes, and significantly improving the accuracy and robustness of object matching in complex scenarios.

[0023] Optionally, in one embodiment of this application, the candidate object acquisition module includes: a name acquisition unit, used to acquire the name of the target object; and an object query unit, used to query at least one candidate object that meets the preset name conditions in the preset database using the name as an index.

[0024] Through the above technical solution, the embodiments of this application can filter out several objects with completely identical names or those that are determined to be similar by the algorithm as candidate objects in the database, thereby narrowing the scope for subsequent similarity matching.

[0025] Optionally, in one embodiment of this application, the similarity calculation module includes: a processing unit, configured to perform semantic vectorization processing on the attribute set of the target object and the attribute set of the at least one candidate object based on a preset language model, so as to obtain the attribute vector set of the target object and the attribute vector set of the at least one candidate object respectively; and a comparison unit, configured to compare the attribute vector set of the target object and the attribute vector set of the at least one candidate object, and calculate the similarity between the target object and the at least one candidate object based on the comparison result.

[0026] Through the above technical solution, the embodiments of this application can first perform quantization processing of the attribute set based on the large language model, and then compare the quantified attribute set to realize the similarity calculation of the target and the candidate object, effectively reducing the problem of deviation in similarity calculation due to the inability of attributes to correspond.

[0027] Optionally, in one embodiment of this application, it further includes: a target object determination module, configured to determine a new target object from multiple publicly available target information sources after constructing the attribute set of the target object; and a first target object attribute set update module, configured to extract the attributes of the new target object based on the context information of the new target object, and update the attribute set using the attributes of the new target object.

[0028] Through the above technical solution, the embodiments of this application can determine the target object from several publicly available information sources, then extract the attributes of the target object, and update the attribute set of the target object, thereby making the attribute set of the target object more complete and increasing the matching accuracy between the target object and the candidate objects in the database.

[0029] Optionally, in one embodiment of this application, it further includes: an associated object acquisition module, configured to acquire the association relationship of the target object after constructing the attribute set of the target object, and obtain at least one associated object based on the association relationship; an associated object attribute set construction module, configured to acquire the attributes of the at least one associated object, and construct the attribute set of the at least one associated object based on the attributes of the at least one associated object; and an associated object attribute set update module, configured to update the attribute set of the target object based on the attribute set of the at least one associated object.

[0030] Through the above technical solution, the embodiments of this application can update the attribute set of the target object based on the attribute set of the associated object, further improve the attribute set, make it contain more valuable information, thereby improving the accuracy and comprehensiveness of subsequent matching with candidate objects in the database.

[0031] Optionally, in one embodiment of this application, the associated object acquisition module includes: a graph structure acquisition unit, configured to acquire a graph structure containing the target object, the graph structure being constructed based on the context information of the target object and used to describe the association relationship between the target object and the at least one associated object; and an associated object determination unit, configured to determine the at least one associated object based on the graph structure.

[0032] Through the above technical solution, the embodiments of this application can obtain related objects based on the obtained graph structure containing the target object, thereby realizing the mining of multi-degree relationships and the revelation of implicit relationships; the technology of obtaining related objects through graph structure is significantly better than traditional association analysis methods in multiple dimensions.

[0033] Optionally, in one embodiment of this application, it further includes: a secondary associated object acquisition module, configured to acquire at least one secondary associated object for each of the at least one associated objects after constructing the attribute set of the at least one associated object; a secondary associated object attribute set construction module, configured to construct the attribute set of the at least one secondary associated object; and a second target object attribute set update module, configured to update the attribute set of the target object based on the attribute set of the at least one secondary associated object.

[0034] Through the above technical solutions, the embodiments of this application can construct a composite attribute set during the object matching process, organically integrating the object's own attributes and relationships. Unlike traditional methods that rely solely on a fixed single attribute for matching, this approach achieves dynamic integration of multi-source data, transforming implicit relationships between objects into computable attribute dimensions. This effectively solves the matching limitations caused by name ambiguity and missing attributes. By constructing a composite feature space that includes semantic and structural relationships, the accuracy and robustness of object matching in complex scenarios are significantly improved.

[0035] A third aspect of this application provides a terminal device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the context-based object matching method as described in the above embodiments.

[0036] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described context-based object matching method.

[0037] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0038] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0039] Figure 1 This is a flowchart of a context-based object matching method provided according to an embodiment of this application;

[0040] Figure 2 This is a flowchart illustrating an object matching method according to a specific embodiment of this application;

[0041] Figure 3 This is a schematic diagram of the structure of a context-based object matching device according to an embodiment of this application;

[0042] Figure 4 This is a block diagram illustrating the internal structure of the terminal device provided in the embodiments of this application. Detailed Implementation

[0043] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0044] The following description, with reference to the accompanying drawings, outlines a context-based object matching method, apparatus, terminal device, and storage medium according to embodiments of this application. Addressing the issues raised in the background section regarding existing technologies, which rely on fixed attributes, lack sufficient context utilization, and have insufficient associative reasoning capabilities, leading to limitations in object identification and alignment within multi-source heterogeneous data, resulting in matching limitations, high false positive rates, and incomplete information completion. This application provides a context-based object matching method. In this method, attributes of the target object are first extracted based on the context information of the target object in the information source, and an attribute set of the target object is constructed. Then, candidate objects are obtained from the database, along with their attribute sets. Similarity matching is performed between the attribute set of the target object and the attribute set of the candidate objects to obtain the similarity between the target object and the candidate objects. Finally, a match is determined based on the similarity score. This method constructs a composite attribute set during object matching, organically integrating the object's own attributes and relationships. Unlike traditional methods that rely solely on fixed single attributes, this method dynamically integrates multi-source data, transforming implicit relationships between objects into computable attribute dimensions. This effectively avoids matching limitations caused by name ambiguity and missing attributes. By constructing a composite feature space encompassing semantic and structural relationships, it significantly improves the accuracy and robustness of object matching in complex scenarios. Therefore, it solves the problems of limited matching, high misjudgment rates, and incomplete information completion in object recognition and alignment of multi-source heterogeneous data, which arise from existing technologies' reliance on fixed attributes, insufficient context utilization, and inadequate association reasoning capabilities.

[0045] With the continuous growth of data volume and the increasing diversification of its sources, accurately identifying and matching the same object across different data sources has become a critical problem that urgently needs to be solved. Existing technologies face numerous insurmountable challenges in addressing this issue.

[0046] On the one hand, the names of the same object appear in extremely complex forms across different data sources. Inconsistencies exist, such as aliases and abbreviations, like "University A" and "A"; ambiguous names also arise, with common names like "Zhang San" potentially corresponding to numerous different individuals. This makes traditional object matching methods based solely on name highly prone to errors. On the other hand, when contextual information is insufficient, existing technologies lack effective means to infer object identity through indirect association. For example, when only a limited amount of information describes someone's participation in a project, but their specific identity is unclear, it's difficult to accurately determine their true identity using other relevant information. Furthermore, existing technologies are insufficient in quantifying the degree of matching between object attributes and database records, often only allowing for simple comparisons and failing to comprehensively consider multiple factors, resulting in low accuracy and reliability of matching results. These problems severely limit the effective integration and utilization of data, significantly reducing the accuracy of data analysis and failing to meet the ever-increasing business demands. Therefore, this application proposes a context-based object matching method to overcome the aforementioned technical deficiencies.

[0047] Specifically, Figure 1 This is a flowchart illustrating a context-based object matching method provided in an embodiment of this application.

[0048] like Figure 1 As shown, this context-based object matching method includes the following steps:

[0049] In step S101, based on the context information of the target object in the information source, at least one attribute of the target object is extracted, and an attribute set of the target object is constructed according to the at least one attribute of the target object.

[0050] In this context, the attributes of the target object can be understood as variables or data items that describe the characteristics of the object.

[0051] like Figure 2 As shown, embodiments of this application can determine the target object to be matched from a given source of information data. Specifically, the source of information can be a text file or other form of data file containing the target object. Figure 2In this example, the given information source is a text file, such as a news report. First, the target object to be matched is determined; in this embodiment, the target object is the person Wang Wu. After identifying Wang Wu as the target object, natural language processing techniques are used to obtain its contextual attributes. Specifically, Named Entity Recognition (NER) technology is used to analyze the text "56-year-old Wang Wu is a professor and doctoral supervisor in the Computer Science Department of University A in City A," accurately identifying "Wang Wu" (person's name), "56 years old" (age), "University A in City A" (institution name), and "professor and doctoral supervisor in the Computer Science Department" (occupation). Further, through keyword matching, in the text "He published a paper titled ...... in August 2024," the paper information is located based on the keyword "published," and "August 2024" (publication time) is extracted using a date recognition model. Finally, dependency parsing is used to analyze the syntactic structure of "works at University A in City A," clarifying that "University A in City A" is the object of "works at," thus accurately extracting the workplace attribute. The age, workplace, occupation, publication time, and other relevant attributes extracted through technical means were compiled into a set of attributes for Wang Wu.

[0052] The embodiments of this application can extract the attributes of the target object, construct the attribute set of the target object, and then perform similarity matching between the attribute set of the target object and the attribute set of the candidate object to calculate the similarity of the candidate object.

[0053] In step S102, at least one candidate object is obtained from a preset database.

[0054] In the embodiments of this application, such as Figure 2 As shown, the database can be a pre-built talent details database, either locally or in the cloud. This database stores data on several objects and their corresponding attributes (e.g., attributes of Wang Wu, Li Si, and Zhang San). Because of the existence of duplicate names in reality, the "Wang Wu" in the talent details database is not necessarily the same as the target object "Wang Wu". Therefore, it is necessary to quickly filter out several candidate objects that may match the target object "Wang Wu" from the talent details database, laying the foundation for subsequent similarity matching.

[0055] The embodiments of this application can obtain candidate objects based on a preset database, and then perform similarity matching between the target object and the candidate objects based on the obtained candidate objects.

[0056] In step S103, the attribute set of at least one candidate object in the preset database is obtained, and the similarity matching between the attribute set of the target object and the attribute set of the candidate object is performed to obtain the similarity between the target object and at least one candidate object.

[0057] like Figure 2 As shown, in this embodiment, after obtaining the attribute set corresponding to the target object Wang Wu, the set is matched with the attribute data of the aforementioned candidate objects in the database, and the candidate object with the highest similarity is selected as the matching object. Alternatively, based on specific needs, several candidate objects can be selected as matching objects based on similarity, for example, the three candidate objects with the highest similarity can be selected as the matching candidates. Regarding the similarity matching algorithm, in this embodiment, it can be performed by customizing a similarity algorithm or using an existing general-purpose large language model.

[0058] The custom similarity algorithm can employ a multi-dimensional calculation strategy. Specifically, firstly, the attribute set of the target object, Wang Wu, and the attribute set of candidate objects in the database are structured. Attributes can be stored using key-value pairs ("attribute-attribute value"), resulting in a key-value pair set such as: {"Institution": "University A", "Occupation": "Professor", "Publication Time": "August 2024"}. Next, multi-dimensional similarity calculations are performed. For textual attributes, the edit distance algorithm is used to quantify differences. Taking "University A" and "A" as an example, character-level similarity is calculated; the smaller the distance, the higher the similarity. For numerical attributes, normalization is first applied to map the values ​​to the [0,1] interval, and then the absolute difference is calculated. Finally, a weighting mechanism can be introduced, assigning weights to different attributes according to scenario requirements, such as a weight of 0.3 for the institution attribute and 0.2 for the occupation attribute. Then, the weighted sum of the similarities of each attribute is calculated to obtain the final similarity value.

[0059] The embodiments of this application can calculate the similarity between the target object and the candidate object, and further achieve matching with the target object based on the similarity.

[0060] In step S104, based on similarity, the matching result between the target object and at least one candidate object is determined.

[0061] As one feasible approach, this embodiment of the application can set a similarity threshold, discarding all candidate objects below the threshold. Among the candidate objects exceeding the threshold, based on specific needs, the object with the highest similarity or several objects with the highest similarity are selected as the matching object. If all candidate objects are discarded, it is determined that no candidate object is matched. This indicates that there is no candidate object matching the target object in the database. The above scheme effectively ensures the reliability and efficiency of matching through threshold filtering and priority selection mechanisms.

[0062] Optionally, in one embodiment of this application, obtaining candidate objects from a preset database includes: obtaining the name of the target object; and querying the preset database for at least one candidate object that meets the preset name conditions, using the name as an index.

[0063] One possible approach is to retrieve candidate objects from a database, which may include:

[0064] Based on the name of the target object, retrieve several candidate objects with the same or similar names from the database.

[0065] In this embodiment, specifically, "identical" means that the character sequence of the names is completely identical. For example, if the target object name is "Wang Wu," then any object in the database whose name field contains "Wang Wu" satisfies the "identical" condition. "Similar" is determined by a quantification algorithm, using, for example, the edit distance (Levenshtein distance) to measure name differences. This distance represents the minimum number of single-character insertion, deletion, and replacement operations required to transform one string into another. A distance threshold is set based on the size of the filtering range. Preferably, in this embodiment, the distance threshold is set to 1. If the edit distance between the target object name "Wang Wu" and the name of an object in the database is less than or equal to the threshold, then it is determined to be "similar." For example, if the edit distance between "Wang Wu" or "Wang Wu Wu" and the target object name "Wang Wu" is 1, which is less than or equal to the distance threshold, then "Wang Wu" or "Wang Wu Wu" is determined to be similar to the target object and is therefore a candidate object.

[0066] Through the above technical solution, the embodiments of this application can filter out several objects with completely identical names or those that are determined to be similar by the algorithm as candidate objects in the database, thereby narrowing the scope for subsequent similarity matching.

[0067] Optionally, in one embodiment of this application, similarity matching is performed on the attribute set of the target object and the attribute set of the candidate object to obtain the similarity between the target object and at least one candidate object, including: semantically vectorizing the attribute set of the target object and the attribute set of at least one candidate object based on a preset language model to obtain the attribute vector set of the target object and the attribute vector set of at least one candidate object respectively; comparing the attribute vector set of the target object with the attribute vector set of at least one candidate object, and calculating the similarity between the target object and at least one candidate object based on the comparison result.

[0068] In this embodiment, because the attribute categories and names of the target object obtained in the context of the information source may not all correspond one-to-one with the attribute categories and names of candidate objects in the database, there may be differences in attribute names such as "job title" and "company". In this case, the matching using the above-mentioned custom similarity algorithm may result in errors in similarity calculation due to mismatched attributes. Therefore, to avoid the above problems, this embodiment uses a general large language model for object similarity matching.

[0069] Unlike the structuring performed before the custom similarity algorithm described above, the structuring using a general large-scale model only records attribute values. The structured result is in the form of: {“Appointed at University A”, “Professor”, “Published in August 2024”}. The structured attribute value set is then input into a large language model. Specifically, BERT, GPT-4, Qwen, or other large language models can be used. The model first segments each attribute value into words and transforms it into an input sequence that the model can process. Then, it uses a multi-layer neural network to semantically encode these sequences, generating corresponding high-dimensional semantic vectors. Preferably, “Appointed at University A” is transformed into a 512-dimensional vector containing semantic information after model processing, and the same applies to attribute values ​​such as “Professor”. All attribute vectors of the target object are then aggregated. In this embodiment, average pooling or weighted summation based on an attention mechanism can be used to form a comprehensive attribute vector of the target object. The attribute value set of the candidate objects is processed in the same way to obtain the comprehensive attribute vector of the candidate objects.

[0070] In the average pooling process, it is first assumed that the target object has n attribute vectors, each with a dimension of d. For example, when processing person information, there may be attributes such as age, occupation, and work location. Each attribute is transformed into a d-dimensional vector by the large language model. Next, for the first dimension of all attribute vectors, their values ​​are summed; then the same operation is performed on the second dimension, and so on, until all d dimensions are summed. Finally, the summation result for each dimension is divided by the number of attribute vectors n to obtain the average value for each dimension. These average values ​​constitute the final composite attribute vector.

[0071] Alternatively, in the attention-based weighted summation process, firstly, for each attribute vector of the target or candidate object, a dot product is performed with a learnable query vector to obtain an attention score. This query vector is learned by the model during training and is used to capture the semantic relationships between different attributes. Next, the attention score is normalized using the softmax function, transforming it into a probability distribution to ensure that the sum of the weights of all attribute vectors is 1. Finally, each attribute vector is multiplied by its corresponding weight, and all results are summed to obtain the final comprehensive attribute vector.

[0072] After obtaining the comprehensive attribute vector through any of the above methods, the cosine similarity formula is used to calculate the similarity between the comprehensive attribute vector of the target object and the comprehensive attribute vector of the candidate object.

[0073] Through the above technical solution, the embodiments of this application can first perform quantization processing of the attribute set based on the large language model, and then compare the quantified attribute set to realize the similarity calculation of the target and the candidate object, effectively reducing the problem of deviation in similarity calculation due to the inability of attributes to correspond.

[0074] Optionally, in one embodiment of this application, after constructing the attribute set of the target object, the method further includes: determining a new target object from multiple publicly available target information sources; extracting the attributes of the new target object based on the context information of the new target object; and updating the attribute set using the attributes of the new target object.

[0075] In this embodiment, the source of the attribute set of the target object can be the context of the information source where the target object resides. Furthermore, to expand and supplement the attribute set of the target object and increase the accuracy of its matching with candidate objects in the database, it is necessary to introduce other publicly available information sources. Specifically, publicly available information sources can be searchable data on the Internet or locally available data.

[0076] The method for identifying target objects from a vast number of publicly available information sources is the same as the matching method for candidate objects described above. First, using the name of the target object, we search for potential objects with the same or similar names in the publicly available information sources. Then, we construct attribute sets for these potential objects using the same method as for constructing the attribute set for the target object, which will not be repeated here. Next, we use the same similarity matching algorithm as described above for matching, which will not be repeated here either. Potential objects from publicly available information sources that meet the requirements are marked as target objects in the publicly available information sources, and their attribute sets are added to the constructed attribute set of the target object.

[0077] Through the above technical solution, the embodiments of this application can determine the target object from several publicly available information sources, then extract the attributes of the target object, and update the attribute set of the target object, thereby making the attribute set of the target object more complete and increasing the matching accuracy between the target object and the candidate objects in the database.

[0078] Optionally, in one embodiment of this application, after constructing the attribute set of the target object, the method further includes: obtaining the association relationship of the target object, and obtaining at least one associated object based on the association relationship; obtaining the attributes of at least one associated object, and constructing the attribute set of at least one associated object based on the attributes of at least one associated object; and updating the attribute set of the target object based on the attribute set of at least one associated object.

[0079] In this embodiment, within the context of the information source, in addition to obtaining the attribute values ​​of the target object, the association between the target object and other objects is also obtained. For example, in the information text "Wang Wu's student Li Si participated in this competition...", it can be deduced that Li Si is Wang Wu's student. In other words, Li Si is an associated object of the target object. Further, the associated object Li Si can be regarded as a new target object, and according to the target attribute set acquisition and similarity matching process described above, Li Si's complete attribute set is obtained. Then, using some attributes from Li Si's attribute set, the attribute set of the original target object Wang Wu is supplemented and updated. Specifically, when Li Si's attribute set records that his current school is University A, while the original attribute set of the target object Wang Wu does not have the attribute of his employing school, it is reasonable to supplement University A, which Li Si attends, as Wang Wu's employing school. This is because in the teacher-student relationship, the school attended by the student is often associated with the school where the teacher is employed.

[0080] Through the above technical solution, the embodiments of this application can update the attribute set of the target object based on the attribute set of the associated object, further improve the attribute set, make it contain more valuable information, thereby improving the accuracy and comprehensiveness of subsequent matching with candidate objects in the database.

[0081] Optionally, in one embodiment of this application, obtaining the association relationship of the target object and obtaining at least one associated object based on the association relationship includes: obtaining a graph structure containing the target object, the graph structure being constructed based on the context information of the target object and used to describe the association relationship between the target object and at least one associated object; and determining at least one associated object based on the graph structure.

[0082] As one feasible approach, this application embodiment, after obtaining the attribute set of the target object based on the context information of the target object in the information source, can construct a graph structure containing the target object based on the context information. Preferably, firstly, from the context text of the target object, using named object recognition and relation extraction algorithms in natural language processing technology, all objects and their relationships, as well as the relationships between them, are extracted. Specifically, keywords such as "guide students" need to be identified, and then the associated object Li Si of the target object Wang Wu can be generated. Based on this, each object is treated as a node, and the relationship is treated as an edge, forming a graph structure. Nodes record the object name and some basic attributes, such as "Wang Wu - Professor", and edges are labeled with relationship type and association strength, such as the "guide students" relationship and weight.

[0083] After constructing the graph structure, the first step is to use the target node (Wang Wu node) as the starting point and search for directly related objects using algorithms such as breadth-first search. For example, weak associations can be filtered based on edge weights to ensure that valid related objects are obtained. Figure 2As shown, all associated objects constitute the associated set of the target object Wang Wu.

[0084] Through the above technical solution, the embodiments of this application can obtain related objects based on the obtained graph structure containing the target object, thereby realizing the mining of multi-degree relationships and the revelation of implicit relationships; the technology of obtaining related objects through graph structure is significantly better than traditional association analysis methods in multiple dimensions.

[0085] Optionally, in one embodiment of this application, after constructing the attribute set of at least one associated object, the method further includes: obtaining at least one secondary associated object for each of the at least one associated object; constructing the attribute set of at least one secondary associated object; and updating the attribute set of the target object based on the attribute set of the at least one secondary associated object.

[0086] In this embodiment, in the graph structure constructed above, the directly related object nodes of the target object node are the nodes directly associated with the target object node. To expand the range of related objects of the target node Wang Wu, based on the breadth-first search for directly related objects, a further search can be conducted to find secondary related objects. Secondary related objects refer to the directly related objects of related objects, such as Zhang San, a classmate of Li Si, Wang Wu's student. Furthermore, not limited to searching for secondary related objects, the traversal depth can be configured according to actual needs, and a breadth-first search can be used to find multi-level indirect related objects layer by layer. All traversed indirect related objects are used to construct a related object set, and the attribute set of the target object is supplemented and updated using the attributes of all related objects in the related object set.

[0087] This application's embodiments can also use a graph-based retrieval enhancement generation model, integrating knowledge graphs with retrieval enhancement generation, to model the relationships between objects through graph structures. For example... Figure 2 As shown, relationships are found iteratively in a chain-like manner based on the knowledge graph.

[0088] Through the above technical solutions, the embodiments of this application can construct a composite attribute set during the object matching process, organically integrating the object's own attributes and relationships. Unlike traditional methods that rely solely on a fixed single attribute for matching, this approach achieves dynamic integration of multi-source data, transforming implicit relationships between objects into computable attribute dimensions. This effectively solves the matching limitations caused by name ambiguity and missing attributes. By constructing a composite feature space that includes semantic and structural relationships, the accuracy and robustness of object matching in complex scenarios are significantly improved.

[0089] The context-based object matching method proposed in this application effectively overcomes the shortcomings of existing technologies and brings significant benefits in many aspects. Regarding attribute integration, by extracting contextual information of the target object from the text, such as institution, project, and research field, and supplementing it with additional attributes such as address and degree using public data sources, a composite attribute set containing its own attributes and related relationships is constructed. This innovation breaks through the limitations of traditional methods that rely solely on fixed attributes, greatly enriching the feature description of the object and significantly improving the accuracy of matching. Taking academic research scenarios as an example, matching researchers solely by name is prone to errors, but by comprehensively considering information such as their institution, projects, and research fields using the method of this invention, the identity can be determined more accurately. Regarding association reasoning and attribute completion, in cases of insufficient contextual information, chain-like association reasoning is performed using known object relationships. Through multiple searches and associations, missing attributes of the object are gradually supplemented, completing iterative attribute completion. This allows for the inference of more complete object information through related information even when some information is missing, reducing the misjudgment rate caused by missing information. For example, in social network data analysis, if some information about a user is missing, the missing information can be supplemented by their relationships with other known users, such as jointly participated activities or friend relationships, leading to a more accurate user profile. In similarity matching, a large model is used to perform comprehensive similarity calculations based on multiple attributes of objects, achieving multi-dimensional similarity matching. Furthermore, the large model is used for classification and similarity analysis to identify potential connections between objects. This makes the matching process more intelligent and comprehensive, better able to handle complex semantics and relationships, and further improves the accuracy and reliability of the matching.

[0090] According to the context-based object matching method proposed in this application, the target object's attributes are first extracted based on the context information of the target object in the information source, and an attribute set of the target object is constructed. Then, candidate objects are obtained from the database, along with their attribute sets. Similarity matching is performed between the target object's attribute set and the candidate object's attribute set to obtain the similarity between the target object and the candidate object. Finally, the similarity is used to determine whether the target object and the candidate object match. This method constructs a composite attribute set during object matching, organically integrating the object's own attributes and relationships. Unlike traditional methods that rely solely on fixed single attributes, this method dynamically integrates multi-source data, transforming implicit relationships between objects into computable attribute dimensions, effectively avoiding matching limitations caused by name ambiguity and missing attributes. By constructing a composite feature space containing semantic and structural relationships, the accuracy and robustness of object matching in complex scenarios are significantly improved. Therefore, this method solves the problems in related technologies, such as limited matching, high misjudgment rates, and incomplete information completion, which arise from existing technologies relying on fixed attributes, insufficient context utilization, and inadequate association reasoning capabilities when processing object recognition and alignment in multi-source heterogeneous data.

[0091] Next, refer to the appendix. Figure 3 This application describes a context-based object matching apparatus according to embodiments thereof.

[0092] Figure 3 This is a block diagram of a context-based object matching device according to an embodiment of this application.

[0093] like Figure 3 As shown, the context-based object matching device 10 includes: a target object attribute set construction module 100, a candidate object acquisition module 200, a similarity calculation module 300, and a matching module 400.

[0094] The target object attribute set construction module 100 is used to extract at least one attribute of the target object based on the context information of the target object in the information source, and construct the attribute set of the target object based on the at least one attribute of the target object.

[0095] The candidate object acquisition module 200 is used to acquire at least one candidate pair from a preset database.

[0096] The similarity calculation module 300 is used to obtain the attribute set of at least one candidate object in a preset database, perform similarity matching between the attribute set of the target object and the attribute set of the candidate object, and obtain the similarity between the target object and at least one candidate object.

[0097] The matching module 400 is used to determine the matching result between the target object and at least one candidate object based on similarity.

[0098] Optionally, in one embodiment of this application, the candidate object acquisition module includes: a name acquisition unit and an object query unit; wherein, the name acquisition unit is used to acquire the name of the target object; and the object query unit is used to query at least one candidate object that meets the preset name conditions in a preset database using the name as an index.

[0099] Optionally, in one embodiment of this application, the similarity calculation module 300 includes: a processing unit and a comparison unit; wherein, the processing unit is used to perform semantic vectorization processing on the attribute set of the target object and the attribute set of at least one candidate object based on a preset language model, so as to obtain the attribute vector set of the target object and the attribute vector set of at least one candidate object respectively; the comparison unit is used to compare the attribute vector set of the target object and the attribute vector set of at least one candidate object, and calculate the similarity between the target object and at least one candidate object based on the comparison result.

[0100] Optionally, in one embodiment of this application, it further includes: a target object determination module and a first target object attribute set update module; wherein, the target object determination module is used to determine a new target object from multiple publicly available target information sources after constructing the attribute set of the target object; the first target object attribute set update module is used to extract the attributes of the new target object based on the context information of the new target object, and update the attribute set using the attributes of the new target object.

[0101] Optionally, in one embodiment of this application, it further includes: an associated object acquisition module, an associated object attribute set construction module, and an associated object attribute set update module; wherein, the associated object acquisition module is used to acquire the association relationship of the target object after constructing the attribute set of the target object, and obtain at least one associated object based on the association relationship; the associated object attribute set construction module is used to acquire the attributes of at least one associated object, and construct the attribute set of at least one associated object based on the attributes of at least one associated object; the associated object attribute set update module is used to update the attribute set of the target object based on the attribute set of at least one associated object.

[0102] Optionally, in one embodiment of this application, the associated object acquisition module includes: a graph structure acquisition unit and an associated object determination unit; wherein, the graph structure acquisition unit is used to acquire a graph structure containing the target object, the graph structure is constructed based on the context information of the target object, and is used to describe the association relationship between the target object and at least one associated object; the associated object determination unit is used to determine at least one associated object based on the graph structure.

[0103] Optionally, in one embodiment of this application, it further includes: a secondary associated object acquisition module, a secondary associated object attribute set construction module, and a second target object attribute set update module; the secondary associated object acquisition module is used to acquire at least one secondary associated object for each of the at least one associated object after constructing the attribute set of at least one associated object; the secondary associated object attribute set construction module is used to construct the attribute set of at least one secondary associated object; and the second target object attribute set update module is used to update the attribute set of the target object based on the attribute set of at least one secondary associated object.

[0104] It should be noted that the foregoing explanation of the context-based object matching method embodiment also applies to the context-based object matching device of this embodiment, and will not be repeated here.

[0105] According to the context-based object matching device proposed in this application, the target object's attributes are first extracted based on the context information of the target object in the information source, and an attribute set of the target object is constructed. Then, candidate objects are obtained from the database, along with their attribute sets. Similarity matching is performed between the target object's attribute set and the candidate object's attribute set to obtain the similarity between the target object and the candidate object. Finally, the similarity is used to determine whether the target object and the candidate object match. This method constructs a composite attribute set during object matching, organically integrating the object's own attributes and relationships. Unlike traditional methods that rely solely on fixed single attributes, this method dynamically integrates multi-source data, transforming implicit relationships between objects into computable attribute dimensions, effectively avoiding matching limitations caused by name ambiguity and missing attributes. By constructing a composite feature space containing semantic and structural relationships, the accuracy and robustness of object matching in complex scenarios are significantly improved. Therefore, this solves the problems in related technologies, such as matching limitations, high misjudgment rates, and incomplete information completion, which arise from existing technologies relying on fixed attributes, insufficient context utilization, and insufficient association reasoning capabilities when processing object recognition and alignment in multi-source heterogeneous data.

[0106] Figure 4 A schematic diagram of the structure of a terminal device provided in an embodiment of this application. The terminal device may include:

[0107] The processor, memory, network interface, display, and temperature sensor are connected via a system bus.

[0108] The terminal device's processor provides computing and control capabilities; its memory includes non-volatile storage media and internal memory; the non-volatile storage media stores the operating system and computer programs; the internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media; the terminal device's network interface is used to communicate with external terminals via a network; the computer program, when executed by the processor, implements a context-based object matching method. The terminal device's display screen can be an LCD screen or an e-ink display screen, and its temperature sensor is pre-installed internally to detect the operating temperature of the internal components.

[0109] Those skilled in the art will understand that Figure 4 The schematic diagram shown is only a partial structural diagram related to the present invention and does not constitute a limitation on the terminal device to which the present invention is applied. The specific terminal device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0110] In one specific embodiment, the terminal device includes a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors. The one or more programs contain instructions for performing the following operations:

[0111] Based on the contextual information of the target object in the information source, extract the attributes of the target object and construct the attribute set of the target object;

[0112] Retrieve candidate objects from the database;

[0113] Obtain the attribute set of the candidate object in the database, perform similarity matching between the attribute set of the target object and the attribute set of the candidate object, and obtain the similarity between the target object and the candidate object;

[0114] Based on similarity, it is determined whether the target object matches the candidate object.

[0115] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0116] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0117] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

[0118] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described context-based object matching method.

[0119] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0120] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0121] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0122] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0123] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or more of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0124] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0125] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0126] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A context-based object matching method, characterized in that, Includes the following steps: Based on the contextual information of the target object in the text information source, extract at least one attribute of the target object, and construct the attribute set of the target object according to the at least one attribute of the target object; Retrieve at least one candidate object from the preset database; Obtain the attribute set of at least one candidate object in the preset database, perform similarity matching between the attribute set of the target object and the attribute set of the candidate object, and obtain the similarity between the target object and at least one candidate object; Based on the similarity, a matching result is determined between the target object and the at least one candidate object; After constructing the attribute set of the target object, the following is also included: Obtain the association relationships of the target object, and obtain at least one associated object based on the association relationships; Obtain the attributes of the at least one associated object, and construct an attribute set for the at least one associated object based on the attributes of the at least one associated object; Update the attribute set of the target object based on the attribute set of the at least one associated object; The step of obtaining the association relationship of the target object and obtaining at least one associated object based on the association relationship includes: Obtain a graph structure containing the target object, the graph structure being constructed based on the context information of the target object and used to describe the association relationship between the target object and the at least one associated object; Based on the graph structure, the at least one associated object is determined.

2. The method according to claim 1, characterized in that, The step of obtaining candidate objects from a preset database includes: Obtain the name of the target object; Using the name as an index, at least one candidate object that meets the preset name conditions is queried in the preset database.

3. The method according to claim 1, characterized in that, The step of performing similarity matching between the attribute set of the target object and the attribute set of the candidate objects to obtain the similarity between the target object and at least one candidate object includes: Based on a preset language model, the attribute set of the target object and the attribute set of the at least one candidate object are semantically vectorized to obtain the attribute vector set of the target object and the attribute vector set of the at least one candidate object, respectively. The attribute vector set of the target object is compared with the attribute vector set of the at least one candidate object, and the similarity between the target object and the at least one candidate object is calculated based on the comparison result.

4. The method according to claim 1, characterized in that, After constructing the attribute set of the target object, the following is also included: Identify new target objects from multiple publicly available sources of information. Based on the context information of the new target object, the attributes of the new target object are extracted, and the attribute set is updated using the attributes of the new target object.

5. The method according to claim 1, characterized in that, After constructing the attribute set of the at least one associated object, the method further includes: Obtain at least one secondary associated object for each of the at least one associated objects; Construct the attribute set of the at least one secondary related object; Update the attribute set of the target object based on the attribute set of the at least one secondary associated object.

6. A context-based object matching device, characterized in that, include: The target object attribute set construction module is used to extract at least one attribute of the target object based on the context information of the target object in the text information source, and construct the attribute set of the target object based on the at least one attribute of the target object; The candidate object acquisition module is used to retrieve at least one candidate object from a preset database. The similarity calculation module is used to obtain the attribute set of at least one candidate object in the preset database, perform similarity matching between the attribute set of the target object and the attribute set of the candidate object, and obtain the similarity between the target object and at least one candidate object; A matching module is used to determine the matching result between the target object and the at least one candidate object based on the similarity. It also includes: an associated object acquisition module, used to acquire the association relationship of the target object after constructing the attribute set of the target object, and obtain at least one associated object based on the association relationship; An associated object attribute set construction module is used to obtain the attributes of the at least one associated object and construct the attribute set of the at least one associated object based on the attributes of the at least one associated object; The associated object attribute set update module updates the attribute set of the target object based on the attribute set of the at least one associated object; The associated object acquisition module includes: The graph structure acquisition unit is used to acquire a graph structure containing the target object, wherein the graph structure is constructed based on the context information of the target object and is used to describe the association relationship between the target object and the at least one associated object; The associated object determination unit is used to determine the at least one associated object based on the graph structure.

7. A terminal device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the context-based object matching method as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the context-based object matching method as described in any one of claims 1-5.

Images