Method, apparatus, device, storage medium and program product for data processing

Abstract

According to embodiments of the present disclosure, methods, apparatuses, devices, storage media, and program products for data processing are provided. The method includes: in response to a first object identifier in a plurality of object identifiers matching a reference object identifier in a knowledge graph, determining a correspondence relationship between the first object identifier and the reference object identifier, the correspondence relationship including a hierarchy of a first object corresponding to the first object identifier in the knowledge graph; in response to at least one second object identifier in the plurality of object identifiers not matching the plurality of reference object identifiers, obtaining a derived object for the knowledge graph; determining an association relationship between the derived object and a reference object corresponding to the reference object identifier, the association relationship including a hierarchy of the derived object in the knowledge graph; and providing ranking information, a plurality of objects being associated with the plurality of object identifiers, the order being determined based on at least the hierarchy of the plurality of objects in the knowledge graph and a plurality of multi-dimensional evaluation data corresponding to the plurality of objects.

Description

Technical Field

[0001] The exemplary embodiments disclosed herein generally relate to the field of computers, and particularly to methods, apparatus, devices, computer-readable storage media, and computer program products for data processing. Background Technology

[0002] As e-commerce platforms expand their business, they need to classify, merge, and sort objects on the platform to provide users with effective object recommendation services. However, achieving highly accurate object classification, merging, and sorting is often challenging. Summary of the Invention

[0003] In a first aspect of this disclosure, a method for data processing is provided. The method includes: in response to a first object identifier among a plurality of object identifiers matching a baseline object identifier in a knowledge graph, determining a correspondence between the first object identifier and the baseline object identifier, the correspondence including the hierarchy of the first object corresponding to the first object identifier in the knowledge graph; in response to at least one second object identifier among the plurality of object identifiers not matching the plurality of baseline object identifiers, obtaining derived objects for the knowledge graph, the derived objects being determined based on the aggregation degree of a target identifier cluster corresponding to the at least one second object identifier; determining an association between the derived objects and the baseline objects corresponding to the baseline object identifiers, the association including the hierarchy of the derived objects in the knowledge graph; and providing ranking information indicating an order of a plurality of objects associated with a plurality of object identifiers, the order being determined at least based on the hierarchy of the plurality of objects in the knowledge graph and corresponding multi-dimensional evaluation data of the plurality of objects, the plurality of objects including at least a first object and a derived object.

[0004] In a second aspect of this disclosure, an apparatus for data processing is provided. The apparatus includes: a correspondence determination module configured to determine a correspondence between the first object identifier and the base object identifier in response to a first object identifier among a plurality of object identifiers matching with a base object identifier in a knowledge graph, the correspondence including the hierarchy of the first object corresponding to the first object identifier in the knowledge graph; a derived object acquisition module configured to acquire a derived object for the knowledge graph in response to at least one second object identifier among the plurality of object identifiers not matching with the plurality of base object identifiers, the derived object being determined based on the aggregation degree of a target identifier cluster corresponding to at least one second object identifier; an association determination module configured to determine an association between the derived object and a base object corresponding to the base object identifier, the association including the hierarchy of the derived object in the knowledge graph; and a ranking information providing module configured to provide ranking information indicating the order of a plurality of objects associated with a plurality of object identifiers, the order being determined at least based on the hierarchy of the plurality of objects in the knowledge graph and corresponding multi-dimensional evaluation data of the plurality of objects, the plurality of objects including at least a first object and a derived object.

[0005] In a third aspect of this disclosure, an electronic device is provided. The device includes at least one processor; and at least one memory coupled to the at least one processor and storing instructions for execution by the at least one processor. When executed by the at least one processor, the instructions cause the device to perform the method of the first aspect.

[0006] In a fourth aspect of this disclosure, a computer-readable storage medium is provided. The computer-readable storage medium stores computer-executable instructions that can be executed by a processor to implement the method of the first aspect.

[0007] In a fifth aspect of this disclosure, a computer program product is provided. The computer program product is tangibly stored in a computer storage medium and includes computer-executable instructions that, when executed by a device, cause the device to perform the method of the first aspect.

[0008] It should be understood that the description in this section is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0009] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1A schematic diagram of an example environment in which embodiments of the present disclosure can be implemented is shown; Figure 2 A flowchart illustrating an example process of data processing according to some embodiments of this disclosure is shown; Figure 3 A flowchart illustrating a data processing procedure according to some embodiments of the present disclosure is shown; Figure 4 A block diagram of an apparatus for data processing according to some embodiments of the present disclosure is shown; and Figure 5 A block diagram of an electronic device capable of implementing one or more embodiments of the present disclosure is shown. Detailed Implementation

[0010] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0011] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0012] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0013] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0014] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0015] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0016] It should be noted that the headings of any section / subsection provided herein are not limiting. Various embodiments are described throughout this document, and embodiments of any type may be included under any section / subsection. Furthermore, embodiments described in any section / subsection may be combined in any way with any other embodiments described in the same section / subsection and / or different sections / subsections.

[0017] In the description of embodiments of this disclosure, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The term "some embodiments" should be understood as "at least some embodiments". Other explicit and implicit definitions may also be included below. The terms "first", "second", etc., may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0018] As used in this paper, the term "model" refers to a system that learns the relationship between inputs and outputs from training data, enabling it to generate corresponding outputs for a given input after training. Model generation can be based on machine learning techniques. Deep learning is a machine learning algorithm that uses multiple layers of processing units to process inputs and provide corresponding outputs. In this paper, "model" may also be referred to as a "machine learning model," a "machine learning network," or simply a "network," and these terms are used interchangeably. A model can also include different types of processing units or networks.

[0019] As used in this document, the term "knowledge graph" (KG) refers to a semantic network that may include a structured description of objects, concepts, and the relationships between them. Knowledge graphs typically include, but are not limited to, object-relationship-object or object-attribute-attribute-value pairs expressed as triples, and can be used to support the standardized, structured description and querying of various objects in the real world. In some embodiments, a knowledge graph may include descriptions of different levels of objects, attributes, and synonym relationships between objects.

[0020] As used herein, the term "object" refers to an entity or entry that can be categorized, merged, and sorted. In some embodiments, objects include, but are not limited to, goods, menu items, content entries, and service items on an e-commerce platform. In other cases, an object can be an entity node in a knowledge graph.

[0021] As used herein, the term "object identifier" refers to a string, text, name, or other form of descriptive information used to identify or characterize an object. In some embodiments, an object identifier includes, but is not limited to, the object's name, object's title, object's alias, object's abbreviation, etc.

[0022] As used herein, the term "reference object identifier" refers to an object identifier that has been pre-determined as a standard. In some embodiments, the reference object identifier may be determined based on authoritative standards or reference databases. The object corresponding to the reference object identifier is referred to herein as a "reference object".

[0023] As used in this paper, the term "first object" refers to the object that corresponds to the first object identifier and has been identified as a baseline object in the knowledge graph. The term "baseline object" refers to the object that corresponds to the baseline object identifier and is included in the knowledge graph backbone. The term "derived object" refers to the object that corresponds to an object identifier that does not match multiple baseline object identifiers and is incorporated into the knowledge graph through derivation.

[0024] As used herein, the term "correspondence" refers to the attribution or matching relationship between an object identifier and a base object identifier. In some embodiments, the correspondence may also include the object's hierarchy in the knowledge graph. The term "association" refers to the attribution or association relationship between a derived object and a base object. In some embodiments, the association may also include the derived object's hierarchy in the knowledge graph.

[0025] As used in this document, the term "hierarchy" refers to the level or rank of an object within a knowledge graph. In some embodiments, the hierarchy can be determined based on the hierarchical structure of the knowledge graph, such as a base object and derived objects being at different levels, or different derived objects being at different levels. In some embodiments, the hierarchy can reflect the degree of standardization of an object, with higher levels indicating a higher degree of standardization.

[0026] As used herein, the term "target identifier cluster" refers to a set of at least one object identifier obtained by clustering multiple object identifiers. In some embodiments, the multiple object identifiers within a target identifier cluster may be different variant representations of the same object, or a set of object identifiers with similar semantics.

[0027] As used in this document, the term "aggregation degree" (also known as "aggregation density") refers to a metric used to quantify the dispersion, frequency of occurrence, and coverage of object identifier clusters. In some embodiments, the aggregation degree can be determined based on the information entropy of the object identifier cluster and the number of e-commerce stores associated with the object identifier cluster. The aggregation degree can be used to determine whether an object identifier cluster needs to be derived into an entity in a knowledge graph.

[0028] As used herein, the term "sorting information" refers to information that indicates the arrangement of multiple objects in a predetermined order. In some embodiments, sorting information may be presented in the form of leaderboards, recommendation lists, sorting result data structures, etc.

[0029] As used in this document, "multi-dimensional evaluation data" refers to data used to evaluate an object across multiple dimensions. In some embodiments, multi-dimensional evaluation data may include, but is not limited to, the object's sales data, user interaction data with the object, and rating data from e-commerce stores associated with the object.

[0030] As used herein, the term "similarity" refers to the degree of closeness or similarity in representation between two object identifiers. In some embodiments, the similarity value can be a continuous value between 0 and 1, with a higher value indicating greater similarity between the two object identifiers. In some embodiments, similarity may include at least one of semantic similarity, feature similarity, and textual similarity.

[0031] As used herein, the term "semantic similarity" refers to the degree of similarity between two object identifiers within a semantic space. The term "feature similarity" refers to the similarity calculated based on multi-dimensional attribute features (e.g., ingredient features, cooking method features, taste features, etc.), and in some embodiments, may be calculated based on cosine similarity. The term "textual similarity" refers to the similarity calculated at the text character level of the object identifiers, and in some embodiments, may be calculated based on a modified edit distance.

[0032] As used herein, the term "information entropy" refers to a metric used to measure the uncertainty of an information source. In some embodiments, information entropy may be calculated based on a predetermined formula (e.g., the Shannon entropy formula or any other suitable formula), where a larger information entropy value indicates a more uncertain and discrete information source.

[0033] As used herein, the term "e-commerce store" refers to a store that offers goods or services on an e-commerce platform. In some embodiments, an e-commerce store may include a restaurant on a food delivery platform, a retailer on an e-commerce platform, etc.

[0034] As used herein, the term "dish" refers to food prepared from cooking ingredients using a specific cooking method. The term "dish name" refers to a string or textual expression used to identify a dish; in some embodiments, this may include non-standardized dish names, such as those with marketing embellishments. The term "recipe standard document" refers to a standardized document published by an authoritative organization that regulates the classification, naming, and composition of dishes. In some embodiments, this may include, but is not limited to, the *Chinese Culinary Dictionary*, *Food Classification Standards for Catering Services*, and *Classification and Naming Standards for Chinese Dishes*, but these are merely exemplary and this disclosure does not limit the scope of the document.

[0035] As used in this article, the term "machine learning model" refers to an algorithmic model that can learn a mapping relationship from input to output based on training data. In some embodiments, the machine learning model may include, but is not limited to, neural network models, decision tree models, support vector machine models, etc.

[0036] As used herein, the term "training" refers to the process of adjusting the parameters of a machine learning model using training data so that the model can perform a predetermined task. In some embodiments, training may include supervised training, unsupervised training, semi-supervised training, and reinforcement training.

[0037] As used in this paper, "Reinforcement Learning" (RL) refers to a class of machine learning paradigms in which an agent adjusts its policy based on reward signals received by interacting with its environment. The term "reward function" refers to the function used to quantify the reward obtained by the agent's actions during reinforcement learning. In some embodiments, the reward function may be designed based on human feedback, such as Reinforcement Learning from Human Feedback (RLHF).

[0038] As used in this paper, the term "embedding" refers to a dense vector representation that maps discrete objects to a continuous vector space. In some embodiments, embeddings can be obtained through a trained neural network model (e.g., a language model, a Transformer model, etc.).

[0039] The term "fine-tuning" as used in this article refers to the process of further training a pre-trained model using data specific to that task to adapt it to that task.

[0040] The term "Natural Language Processing" (NLP) as used in this article refers to the technical field of using computers to process, analyze, and understand human language.

[0041] As used in this paper, "clustering" refers to the unsupervised learning process of dividing a set of samples into multiple clusters based on the similarity between samples. In some embodiments, clustering may include, but is not limited to, K-Means clustering, density clustering (e.g., DBSCAN), and hierarchical clustering.

[0042] The term "edit distance" (also known as Levenshtein distance) used in this article refers to the minimum number of editing operations (including insertions, deletions, and replacements) required to transform one string into another. In some embodiments, the edit distance can be improved to reduce the interference of character length differences on similarity calculations.

[0043] As e-commerce platforms expand, the number of objects (such as dishes, products, and content items) on the platform continues to increase, and the naming conventions for object identifiers (such as dish names) are becoming increasingly non-standardized and personalized. Different e-commerce stores may use different object identifiers for the same object. For example, for the dish "stir-fried pork," different stores might use different names such as "farmhouse-style stir-fried pork," "farmhouse-style stir-fried pork," "chili-fried pork," and "Hunan-style stir-fried pork." To provide users with effective object recommendation services and improve the accuracy of object ranking, it is necessary to classify, merge, and sort multiple objects on the platform. However, the large number of objects, non-standardized naming, and rapid object iteration make object classification, merging, and sorting challenging.

[0044] Conventionally, one approach is to classify and merge object identifiers using rule-based matching. However, this method typically relies on manually pre-defined thesaurus, which cannot exhaustively cover all synonym variations. Adding new objects requires continuous updates to the rule base, resulting in high maintenance costs and low classification coverage. Another approach uses natural language processing algorithms to extract features from object identifiers, calculates the similarity between pairs of object identifiers, and determines whether they are the same object based on a fixed threshold. However, this method can only extract shallow textual features from object identifiers, limiting semantic understanding capabilities. Using a fixed similarity threshold for judgment can easily lead to the mismerging of objects that are similar in form but different in substance; for example, it might incorrectly merge "stir-fried pork (pork)" with "stir-fried beef (beef)," which are similar in form but fundamentally different.

[0045] Furthermore, manually constructed knowledge graphs can be used to assist in the classification and merging of objects. However, this method also relies on manual annotation, and the core objects and relationships of the knowledge graph are fixed, making it unsuitable for scenarios with rapid object iteration and numerous regionally specific objects. Consequently, this solution also struggles to automatically attach massive amounts of non-standardized object identifiers to a standardized object system, resulting in low dynamic coverage and difficulty in supporting the precise and personalized requirements for object sorting.

[0046] In view of this, an improved scheme for data processing is proposed. According to this scheme, if a first object identifier among a plurality of object identifiers is determined to match a baseline object identifier in a knowledge graph, a correspondence between the first object identifier and the baseline object identifier is determined. In some embodiments, the correspondence may include the hierarchy of the first object corresponding to the first object identifier in the knowledge graph. Further, if at least one second object identifier among the plurality of object identifiers is determined not to match the plurality of baseline object identifiers, a derived object for the knowledge graph is obtained. In some embodiments, the derived object may be determined based on the aggregation degree of the target identifier cluster corresponding to at least one second object identifier. Further, an association relationship between the derived object and the baseline object corresponding to the baseline object identifier is determined. In some embodiments, the association relationship may include the hierarchy of the derived object in the knowledge graph. Furthermore, ranking information may be provided, which may indicate the order of multiple objects associated with multiple object identifiers, the order being determined at least based on the hierarchy of the multiple objects in the knowledge graph and the corresponding multi-dimensional evaluation data of the multiple objects, the multiple objects including at least the first object and the derived object.

[0047] In this way, matching object identifiers can be efficiently attached to baseline objects in the knowledge graph. Correspondingly, derived objects for the knowledge graph can be dynamically obtained based on the aggregation degree of target identifier clusters formed by unmatched object identifiers. This allows for the formation of a hierarchical knowledge graph that can adaptively expand based on business data. The ranking information determined by the hierarchy and multi-dimensional evaluation data in the knowledge graph can take into account multiple dimensions such as the standardization degree of objects, their actual business performance, and user interaction behavior, thereby improving the accuracy, rationality, and diversity of object ranking.

[0048] Figure 1 A schematic diagram of an example environment 100 according to some scenarios of this disclosure is shown. For example... Figure 1 As shown, example environment 100 may include electronic device 110.

[0049] In some embodiments, the electronic device 110 may run an application 120 that supports data processing. In some embodiments, the application 120 may be an application for presenting sorting information of objects, such as an e-commerce application, a food delivery application, a content distribution application, etc.

[0050] In some embodiments, user 140 can interact with application 120 through electronic device 110. Electronic device 110 can present interface 150 to user 140 through application 120. In some embodiments, interface 150 can present sorting information including multiple objects, such as a ranking list of dishes, a recommended list of products, etc.

[0051] In some embodiments, electronic device 110 can communicate with server 130. In some embodiments, electronic device 110 can send input or requests from user 140 to server 130; server 130 can perform data processing based on the input or request (e.g., object identifier matching based on knowledge graph, acquisition of derived objects, determination of association relationships, and generation of ranking information), and send the obtained ranking information to electronic device 110 for presentation to interface 150 by application 120.

[0052] In some embodiments, electronic device 110 communicates with server 130 to provide services to application 120. Electronic device 110 can be any type of mobile terminal, fixed terminal, or portable terminal, including mobile phones, desktop computers, laptop computers, notebook computers, netbook computers, tablet computers, media computers, multimedia tablets, personal communication system (PCS) devices, personal navigation devices, personal digital assistants (PDAs), audio / video players, digital cameras / camcorders, positioning devices, television receivers, radio receivers, e-book devices, gaming devices, or any combination thereof, including accessories and peripherals of these devices or any combination thereof. In some embodiments, electronic device 110 can also support any type of user-facing interface (such as "wearable" circuitry). Server 130 can be various types of computing systems / servers capable of providing computing power, including but not limited to mainframes, edge computing nodes, computing devices in cloud environments, etc.

[0053] It should be understood that the structure and function of the various elements in environment 100 are described for illustrative purposes only and do not imply any limitation on the scope of this disclosure.

[0054] The examples will continue to be described with reference to the accompanying drawings. In the following description, the examples will be primarily described with respect to electronic device 110. It should be understood that the actions described with respect to electronic device 110 can also be performed by an agent deployed at electronic device 110, or by an agent in collaboration with its server (e.g., server 130).

[0055] For ease of understanding, the following will refer to Figure 2 This disclosure describes the data processing scheme used in this publication. Figure 2A flowchart of an example process 200 for data processing according to some embodiments of the present disclosure is shown.

[0056] In some embodiments, at block 211, electronic device 110 can construct a knowledge graph. The knowledge graph can include multiple entities expressed in a structured manner and the relationships between entities. In some embodiments, the knowledge graph can be constructed based on authoritative standard documents. For example, the knowledge graph can be constructed based on recipe standard documents. Recipe standard documents may include, but are not limited to, the *Chinese Culinary Dictionary*, the *Classification Standard for Food in Catering Services*, and the *Classification and Naming Standard for Chinese Dishes* published by authoritative institutions. Based on recipe standard documents, a standardized set of core objects for dishes can be extracted, common names and niche alternative names can be eliminated, and a unique identifier and definition for each standard dish can be established.

[0057] In some embodiments, a knowledge graph may include relationships between entities expressed in the form of triples. In some embodiments, the triples may include hierarchical triple types, such as <object class, contains, base object>, for example, <"Stir-fry category", contains, "Farmhouse stir-fried pork">, to indicate hierarchical classification relationships of objects. The triples may include attribute description triple types, such as <base object, has, attribute value>, which can cover attributes such as cuisine, main / side ingredients, cooking method, flavor, spiciness, and regional affiliation, to indicate the multi-dimensional attribute characteristics of the base object. The triples may include synonymous association triple types, such as <base object, equivalent, synonym>, to indicate the synonymous representation of the base object.

[0058] In some embodiments, the baseline objects in the knowledge graph can be subject to an admission threshold based on authoritative standards, such as including only dishes that are specified in authoritative standard documents and are common within a predetermined geographical area. This provides a unique benchmark for matching subsequent non-standardized object identifiers with the baseline objects in the knowledge graph.

[0059] In some embodiments, at block 212, the electronic device 110 can acquire multiple initial object identifiers and preprocess them. In some embodiments, the multiple initial object identifiers may include, but are not limited to, the names of dishes uploaded by e-commerce stores. Since the names of dishes uploaded by e-commerce stores may have problems such as redundant characters, marketing modifiers, inconsistent character formats, and semantic ambiguity, the multiple initial object identifiers can be preprocessed to extract the core semantics.

[0060] In some embodiments, preprocessing may include a first-level preprocessing and a second-level preprocessing. In some embodiments, the first-level preprocessing may include removing redundant characters from each initial object identifier using regular expressions. For example, removing special symbols, spaces, and marketing modifiers (such as "signboard," "secret recipe," etc.). Redundant characters may also include quantifiers, punctuation marks, consistent traditional / simplified Chinese character formatting, capitalization, etc. Additionally / alternatively, redundant characters may also include invalid data that is not part of the object class, such as business names, promotional slogans, or garbled strings. Through this first-level processing, redundant characters can be removed from multiple initial object identifiers, resulting in preliminarily cleaned object identifiers.

[0061] In some embodiments, the second-layer preprocessing can remove non-entity semantic characters, such as function words and adjectives, from the initially purified object identifiers based on a pre-defined invalid semantic lexicon. Through the second-layer preprocessing, only characters indicating the core semantics of the object are retained, such as core words like ingredients, cooking methods, and flavors. For example, the object identifier "signature farmhouse stir-fried pork with chili peppers" can be simplified to "stir-fried pork with chili peppers" after the second-layer preprocessing. This approach maximizes the preservation of the effective semantics of the object identifiers while eliminating irrelevant and interfering information, thereby improving the accuracy of subsequent matching.

[0062] In some embodiments, after preprocessing, the electronic device 110 can extract multi-dimensional feature vectors of multiple object identifiers. In some embodiments, global semantic feature vectors of object identifiers can be extracted based on a trained pre-trained model (e.g., a text-based pre-trained language model), hereinafter referred to as...

[0063] In some embodiments, the pre-trained model can be a model fine-tuned for the specific dish context, adapting it to the semantic understanding task of dish names. Furthermore, feature vectors for core ingredients, core cooking methods, and core flavors can be extracted from object identifiers based on part-of-speech tagging and keyword matching, as hereinafter referred to as follows:

[0064] In some embodiments, the above multi-dimensional feature vectors can be fused based on the following formula to obtain a comprehensive feature vector for object identification, hereinafter represented as follows: :

[0065] in, , , , For the weighting coefficients, satisfying In some embodiments, the similarity calculation can be set based on the contribution of different dimensions of the object identifier. Greater than , Greater than , Greater than This approach maximizes the weight of the object's food-related features. By enhancing the object's identifier based on its essential characteristics, the accuracy of similarity calculations between the object's identifier and the baseline object's identifier can be improved.

[0066] By preprocessing multiple initial object identifiers, redundant characters and characters without entity semantics can be removed from them, thereby improving the semantic clarity of the object identifiers, reducing noise in subsequent similarity calculations, and thus improving the accuracy of matching between the object identifiers and the baseline object identifiers.

[0067] In some embodiments, at block 213, electronic device 110 may perform matching of object identifiers with reference object identifiers in a knowledge graph based on a plurality of preprocessed object identifiers. In some embodiments, if electronic device 110 determines that a first object identifier among a plurality of object identifiers matches a reference object identifier in the knowledge graph, it determines a correspondence between the first object identifier and the reference object identifier. The correspondence may include the hierarchy of the first object corresponding to the first object identifier in the knowledge graph.

[0068] In some embodiments, the electronic device 110 can determine whether the first object identifier and the reference object identifier match based on the similarity between the first object identifier and the reference object identifier. For example, if the electronic device 110 determines that the similarity between the first object identifier and the reference object identifier meets a similarity threshold, then the first object identifier matches the reference object identifier. In some embodiments, the similarity threshold can be determined based on the category to which the reference object identifier belongs. For example, different similarity thresholds can be used for different food categories. In this way, false matches caused by using a single fixed similarity threshold can be avoided, and the flexibility and accuracy of object identifier matching can be improved.

[0069] In some embodiments, the electronic device 110 may determine the similarity between a first object identifier and a reference object identifier based on at least one of semantic similarity, feature similarity, and textual similarity. In some embodiments, the semantic similarity may be a similarity determined based on a trained first machine learning model, indicating the degree of proximity between the first object identifier and the reference object identifier in the semantic space. In some embodiments, the first machine learning model may be trained based on a reward function, such as reinforcement learning based on human feedback. In some embodiments, the reward function may be determined based on the following formula: Where R is the reward function, α and ε are reward coefficients, AccMatch is the model matching accuracy, and PassCheck is the expert verification pass rate. By iteratively training the first machine learning model based on the above reward function, the first machine learning model can gradually learn the semantic patterns of object identifier matching, thereby improving the accuracy of semantic similarity calculation.

[0070] In some embodiments, the electronic device 110 can fine-tune a first machine learning model through supervised training to enable the first machine learning model to initially learn object matching rules. Then, the electronic device 110 iteratively optimizes the first machine learning model using reinforcement learning based on human feedback. During the supervised training phase, multiple high-quality object identifier matching samples labeled manually can be used as training data. During the reinforcement learning phase, the parameters of the first machine learning model can be adjusted according to the aforementioned reward function to gradually improve the matching accuracy of the first machine learning model.

[0071] In some embodiments, the electronic device 110 may determine feature similarity based on multi-dimensional attribute features associated with a first object identifier. In some embodiments, the multi-dimensional attribute features may include, but are not limited to, food characteristics, cooking method characteristics, and flavor characteristics. In some embodiments, the electronic device 110 may determine feature similarity based on the cosine similarity between the feature vector of the first object identifier and the feature vector of a reference object identifier.

[0072] In some embodiments, the electronic device 110 may determine text similarity based on the proximity of a first object identifier and a reference object identifier at the text character level. In some embodiments, text similarity may be calculated based on an improved edit distance. By improving the edit distance, interference from differences in character length can be eliminated in the similarity calculation.

[0073] In some embodiments, the electronic device 110 may weight the semantic similarity, feature similarity, and text similarity to determine the similarity. In some embodiments, the similarity may be determined based on the following formula: , in For the first object identifier With reference object identifier Similarity between them Represents an object identifier. Represents a base object identifier. For the first object identifier With reference object identifier The semantic similarity between them ranges from 0 to 1. For the first object identifier With reference object identifier The feature similarity between them ranges from 0 to 1. For the first object identifier With reference object identifier The text similarity between the two texts ranges from 0 to 1. Based on the weights, satisfying . This is the deviation correction factor. This is the deviation term between semantic similarity and feature similarity.

[0074] In some embodiments, the base weights can be determined through iterative optimization, for example... It can be 0.45. It can be 0.35. It can be 0.2. The deviation correction coefficient can be determined through iterative optimization, for example... It can be 0.05. This is achieved by introducing a deviation correction factor. and deviation term This can avoid the influence of a single dimension of similarity on the final similarity score, and improve the ability to distinguish between similar but essentially different objects.

[0075] In some embodiments, the electronic device 110 can determine whether the first object identifier and the reference object identifier match based on a comparison of the similarity between the first object identifier and the reference object identifier and a similarity threshold. For example, if the electronic device 110 determines that the similarity is greater than or equal to the similarity threshold, it can determine that the first object identifier matches the reference object identifier and associate the first object identifier with the reference object corresponding to the reference object identifier. Accordingly, the electronic device 110 can determine the level of the first object corresponding to the first object identifier as the level of the reference object in the knowledge graph. In some examples, if the electronic device 110 determines that the similarity is less than the similarity threshold, it can determine that the first object identifier and the reference object identifier do not match. In some embodiments, the electronic device 110 can adjust the similarity threshold according to the category to which the reference object identifier belongs. For example, for "stir-fry," the similarity threshold can be set to 0.82; for "soup," the similarity threshold can be set to 0.78. These are merely exemplary, and this disclosure is not limited in this respect.

[0076] By determining semantic similarity through multi-dimensional similarity and a machine learning model trained using reinforcement learning based on human feedback, the matching between object identifiers and baseline object identifiers can incorporate multiple judgment criteria at the semantic, feature, and text levels, thereby improving matching accuracy. Introducing a bias correction coefficient and a bias term can avoid misjudgments based on single-dimensional similarity and reduce the erroneous merging rate of similar but fundamentally different objects.

[0077] In some embodiments, at block 214, if the electronic device 110 determines that at least one second object identifier among a plurality of object identifiers does not match a plurality of baseline object identifiers, it can obtain a derived object for the knowledge graph. In some embodiments, the derived object may be determined based on the aggregation degree of the target identifier cluster corresponding to at least one second object identifier.

[0078] In some embodiments, the electronic device 110 can determine a target identifier cluster by performing clustering processing on at least one second object identifier. Further, the electronic device 110 can determine the degree of aggregation based at least on the information entropy of the target identifier cluster and the number of at least one e-commerce store, where the number indicates the number of e-commerce stores including objects corresponding to the target identifier cluster. Further, the electronic device 110 obtains derived objects for the knowledge graph based on the degree of aggregation and a degree of aggregation threshold range.

[0079] In some embodiments, clustering processing may include, but is not limited to, K-Means clustering, density clustering (e.g., DBSCAN), and hierarchical clustering. Through clustering processing, at least one second object identifier can be divided into one or more target identifier clusters, wherein multiple object identifiers within each target identifier cluster have similar semantics or expressions.

[0080] In some embodiments, information entropy can be determined based on the following formula: ,in For target identifier cluster Information entropy; For target identifier cluster Internal object identifier variant The proportion; M is the target identifier cluster. The number of object identifier variants within. In some embodiments, information entropy. The larger the target identifier cluster, the stronger the target identifier cluster. The higher the discreteness and uncertainty of the object identifiers within a given range, the better it can represent a target identifier cluster. The degree of diversity across different e-commerce stores.

[0081] In some embodiments, the degree of aggregation can be determined based on the following formula: , in For target identifier cluster The weighted aggregation degree; For target identifier cluster The total frequency of object identifiers within. The total number of samples identified by the object. This represents the maximum information entropy. This is the dispersion correction factor. The e-commerce store coverage coefficient can be compared with the target identifier cluster. The number of associated e-commerce stores is positively correlated. In this way, the aggregation degree can simultaneously reflect the target identifier cluster. The frequency, dispersion, and e-commerce store coverage of the data are used to improve the quantification accuracy of the object's versatility.

[0082] In some embodiments, the aggregation degree threshold range may include a first aggregation degree threshold and a second aggregation degree threshold, wherein the first aggregation degree threshold is greater than the second aggregation degree threshold. Based on the comparison between the aggregation degree and the aggregation degree threshold, the electronic device 110 can determine the hierarchy of the derived object.

[0083] In some embodiments, if the electronic device 110 determines that the aggregation degree is greater than a first aggregation degree threshold, it can identify the derived object as a first-level derived object. In some embodiments, the first-level derived object may belong to the same level as the base object. The electronic device 110 can incorporate the first-level derived object as a new base object into the knowledge graph backbone branch and inherit the attribute system of the base object that has the same hierarchical relationship as the first-level derived object. In this way, frequently occurring objects with cross-regional universality can be automatically derived into objects in the knowledge graph backbone.

[0084] In some embodiments, if the electronic device 110 determines that the aggregation degree is greater than a second aggregation degree threshold and less than a first aggregation degree threshold, it can identify the derived object as a second-level derived object. In some embodiments, the second-level derived object may be a child of a base object. The electronic device 110 can attach the second-level derived object to the base object corresponding to the regional category to which the target identifier cluster belongs. In this way, the diversity of regionally characteristic objects and the hierarchical structure of the map can be balanced.

[0085] In some embodiments, if the electronic device 110 determines that the aggregation degree is less than a second aggregation degree threshold, it can identify the derived object as a predetermined derived object. In some embodiments, the predetermined derived object may be associated with the attributes of an e-commerce store. In some embodiments, the predetermined derived object may only have personalized attribute annotations and may not be included in the backbone of the knowledge graph. In this way, niche objects that appear only in a single e-commerce store or a few e-commerce stores can be included in the knowledge graph as predetermined derived objects without compromising the standardization of the knowledge graph backbone.

[0086] In some embodiments, the electronic device 110 can determine the association between a derived object and a base object corresponding to a base object identifier. In some embodiments, the association may include the hierarchy of the derived object in the knowledge graph. In some embodiments, the hierarchy corresponding to the base object may be higher than the hierarchy corresponding to the derived object. In some embodiments, the hierarchy corresponding to a first-level derived object may be higher than the hierarchy corresponding to a second-level derived object.

[0087] By employing a derived object acquisition mechanism based on aggregation degree, knowledge graphs can be expanded using real-world data without human intervention. This enhances the knowledge graph's adaptability to rapidly iterating objects and improves object activity coverage. Furthermore, by hierarchically classifying derived objects, the knowledge graph maintains a hierarchical structure while accommodating the generality and diversity of different objects.

[0088] In some embodiments, at box 215, the electronic device 110 can autocomplete and differentiate the attributes of multiple objects.

[0089] In some embodiments, for a base object, first-level derived objects, and second-level derived objects, the electronic device 110 can automatically complete attributes based on attribute inheritance rules in a knowledge graph. For example, the electronic device 110 can automatically inherit the attribute system of the base object to derived objects at the same level or belonging to the child level of the base object based on triple relationships in the knowledge graph. In some embodiments, the electronic device 110 can verify the accuracy of the inherited attributes based on the following formula: , in For attribute validation values, To infer attributes With reference attributes The similarity between them. By verifying the accuracy of the inherited attributes, erroneous attributes can be eliminated.

[0090] In some embodiments, the electronic device 110 can determine the attributes of multiple objects based on the attribute information of e-commerce stores associated with multiple object identifiers. In some embodiments, the attribute information of e-commerce stores may include, but is not limited to, at least one of the following: the main cuisine of the e-commerce store, regional tags, merchant qualifications, and user review keywords. For example, for the "stir-fried pork" object: in the scenario where the main cuisine of the corresponding e-commerce store is "Sichuan cuisine," attributes such as "Sichuan cuisine," "medium spicy," and "stir-fried" can be labeled on the object. In the scenario where the main cuisine of the corresponding e-commerce store is "Hunan cuisine," attributes such as "Hunan cuisine," "extra spicy," and "dry-fried" can be labeled on the object. In the scenario where the corresponding e-commerce store is a fast food restaurant, attributes such as "home-style dish," "mildly spicy," and "stir-fried" can be labeled on the object. In this way, the attributes of the same baseline object can be differentiated according to the characteristics of different e-commerce stores, thereby avoiding the homogeneous presentation of the same object in the sorting information of different e-commerce stores.

[0091] In some embodiments, the electronic device 110 can also update the attributes of at least one of a plurality of objects in response to updates to information or review data from e-commerce stores. This ensures the timeliness and accuracy of the object attributes, providing accurate multi-dimensional data support for the generation of subsequent sorting information.

[0092] By automatically completing the attributes of multiple objects and differentially labeling those attributes, the attributes of multiple objects can conform to the standardized rules in the knowledge graph. Furthermore, it can also present the characteristics of objects from different e-commerce stores, thereby avoiding homogenization in object ranking.

[0093] In some embodiments, at block 216, electronic device 110 can determine the ranking score of multiple objects based on the hierarchy of multiple objects in the knowledge graph and the corresponding multi-dimensional evaluation data of multiple objects.

[0094] In some embodiments, multi-dimensional evaluation data may include, but is not limited to, at least one of the following: sales data corresponding to multiple objects, interaction behaviors, and evaluation data corresponding to e-commerce stores. In some embodiments, sales data may include metrics such as the sales volume and sales revenue of the object. Interaction behaviors may include data such as click-through rate, collection rate, and sharing rate corresponding to metrics such as user click behavior, collection behavior, and sharing behavior of the object. Evaluation data may include data corresponding to metrics such as user ratings of the object and the e-commerce store associated with the object, and evaluation text.

[0095] In some embodiments, the ranking scores of multiple objects can be determined based on the following formula: , in The overall ranking score for object F. The standardized score is given for object F, indicating its level in the knowledge graph. In some embodiments, the standardized score of the baseline object can be 1.0, the standardized score of the derived object at the first level can be 0.9, the standardized score of the derived object at the second level can be 0.7, and the standardized score of the predetermined derived object can be 0.5. However, these are merely exemplary and are not intended to limit the scope of the invention. This is a normalized score based on sales data. This is a normalized score based on user ratings in the evaluation data. The popularity score is normalized based on interactive behavior (such as click-through rate and collection rate). This is a normalized score based on reputation data of e-commerce stores. to For the weighting coefficients, satisfying .

[0096] In some embodiments, the level corresponding to the baseline object can be higher than the level corresponding to the derived object. In some embodiments, the weighting coefficients can be weighting coefficients corresponding to the standardized scores of the corresponding levels. Set it higher than other weighting coefficients, for example It can be set to 0.3, and the other weight coefficients can be adaptively adjusted according to the actual scenario. By setting the weight coefficient corresponding to the level to the highest, objects with higher standardization scores can be given higher priority in the sorting, reducing the proportion of non-standardized and low-quality objects in the sorting, thereby avoiding problems such as duplicate display and inaccurate classification in object sorting.

[0097] In some embodiments, at block 217, the electronic device 110 can generate ranking information by arranging multiple objects in descending order of their ranking scores. In some embodiments, the ranking information may be a leaderboard, a recommendation list, or other form of ranking result that includes multiple objects.

[0098] In some embodiments, electronic device 110 can acquire sorting information. Electronic device 110 can then present the acquired sorting information to interface 150 via application 120 to display sorting information including multiple objects to user 140.

[0099] In some embodiments, ranking information can be further integrated with business scenarios such as search recommendations and leaderboards. For example, ranking information can serve as input data for search recommendation results for users. Or, ranking information can serve as input data for leaderboards. In this way, ranking information obtained based on knowledge graphs can be widely applied to various business scenarios, improving the accuracy and applicability of ranking.

[0100] By using the standardized score corresponding to an object's level in the knowledge graph as the core factor in the ranking score, and incorporating sales data, interaction behavior, and evaluation data as auxiliary factors, the resulting ranking information can both consider the object's standardization level and reflect its actual business performance and user preferences. Improving the accuracy, rationality, and diversity of object ranking can enhance user acceptance of the ranking information and improve business metrics such as click-through rate and conversion rate.

[0101] Figure 3 A flowchart of a data processing procedure 300 according to some scenarios of this disclosure is shown. Procedure 300 can be implemented at electronic device 110, at server 130, or jointly implemented by electronic device 110 and server 130. For ease of discussion, the following description assumes that procedure 300 is implemented at electronic device 110. References below... Figure 3 To describe process 300.

[0102] like Figure 3 As shown in block 310, in response to a first object identifier among a plurality of object identifiers being matched with a base object identifier in a knowledge graph, electronic device 110 determines a correspondence between the first object identifier and the base object identifier, the correspondence including the level of the first object corresponding to the first object identifier in the knowledge graph.

[0103] In box 320, if at least one second object identifier among multiple object identifiers of electronic device 110 does not match multiple baseline object identifiers, a derived object for the knowledge graph is obtained, the derived object being determined based on the aggregation degree of the target identifier cluster corresponding to at least one second object identifier.

[0104] In box 330, electronic device 110 determines the association between the derived object and the base object corresponding to the base object identifier, the association including the hierarchy of the derived object in the knowledge graph.

[0105] In box 340, electronic device 110 provides sorting information indicating the order of multiple objects associated with multiple object identifiers. The order is determined at least based on the hierarchy of the multiple objects in the knowledge graph and the corresponding multi-dimensional evaluation data of the multiple objects. The multiple objects include at least a first object and derived objects.

[0106] In some embodiments, a first object identifier is determined to match a reference object identifier by: determining that the first object identifier matches the reference object identifier in response to the similarity between the first object identifier and the reference object identifier meeting a similarity threshold, wherein the similarity threshold is determined based on the category to which the reference object identifier belongs.

[0107] In some embodiments, the similarity is determined based on at least one of semantic similarity, feature similarity, and textual similarity between the first object identifier and the baseline object identifier.

[0108] In some embodiments, the similarity is further determined based on the deviation between semantic similarity and feature similarity to determine the update.

[0109] In some embodiments, semantic similarity is determined based on a first machine learning model trained on a reward function.

[0110] In some embodiments, feature similarity is determined based on multi-dimensional attribute features associated with a first object identifier.

[0111] In some embodiments, multiple object identifiers are obtained based on preprocessing of multiple initial object identifiers, the preprocessing including at least one of the following: removing redundant characters from each initial object identifier; and removing non-entity semantic characters from each initial object identifier based on a preset invalid semantic lexicon.

[0112] In some embodiments, obtaining derived objects for a knowledge graph includes: determining a target identifier cluster by performing clustering processing on at least one second object identifier; determining a degree of aggregation based at least on the information entropy of the target identifier cluster and the number of at least one e-commerce store, the number indicating the number of e-commerce stores including objects corresponding to the target identifier cluster; and obtaining derived objects for the knowledge graph based on the degree of aggregation and a degree of aggregation threshold range.

[0113] In some embodiments, the aggregation degree threshold range includes a first aggregation degree threshold and a second aggregation degree threshold, wherein the first aggregation degree threshold is greater than the second aggregation degree threshold, and obtaining derived objects for the knowledge graph includes: in response to an aggregation degree greater than the first aggregation degree threshold, determining the derived object as a first-level derived object, wherein the first-level derived object belongs to the same level as the base object; in response to an aggregation degree greater than the second aggregation degree threshold and less than the first aggregation degree threshold, determining the derived object as a second-level derived object, wherein the second-level derived object belongs to a child of the base object; or in response to an aggregation degree less than the second aggregation degree threshold, determining the derived object as a predetermined derived object, wherein the predetermined derived object is associated with the attributes of the e-commerce store.

[0114] In some embodiments, the attributes of the second-level derived object are determined based on the attributes of the parent object corresponding to the second-level derived object in the knowledge graph.

[0115] In some embodiments, the multi-dimensional evaluation data includes at least one of the following: sales data of multiple objects, interaction behavior, and evaluation data of e-commerce stores.

[0116] In some embodiments, the hierarchy corresponding to the base object is higher than the hierarchy corresponding to the derived object.

[0117] In some embodiments, the multiple object identifiers include multiple dish names, the multiple objects include multiple dishes corresponding to the multiple dish names, and the baseline object identifiers are determined based on the recipe standard file.

[0118] In some embodiments, the attributes of multiple objects are determined based on the attribute information of the e-commerce stores associated with multiple object identifiers.

[0119] In some embodiments, process 300 further includes updating the attribute of at least one of the plurality of objects in response to an update of information or review data of the e-commerce store.

[0120] In this way, the first matching object identifier can be efficiently attached to the base object in the knowledge graph, and the derived object for the knowledge graph can be obtained based on the aggregation degree of the target identifier cluster formed by the unmatched second object identifier. This allows the knowledge graph to be dynamically expanded based on actual data, and ranking information can be generated based on the hierarchical and multi-dimensional evaluation data of multiple objects, which can improve the accuracy and rationality of object ranking.

[0121] A corresponding apparatus for implementing the above methods or processes is also provided. Figure 4 A block diagram of an apparatus 400 for data processing according to some embodiments of the present disclosure is shown. Apparatus 400 may be implemented as or included in electronic device 110 or server 130. Various modules / components in apparatus 400 may be implemented by hardware, software, firmware, or any combination thereof.

[0122] like Figure 4As shown, the device 400 includes a correspondence determination module 410, configured to determine a correspondence between the first object identifier and the base object identifier in response to a first object identifier among a plurality of object identifiers matching a base object identifier in a knowledge graph. The correspondence includes the level of the first object corresponding to the first object identifier in the knowledge graph. The device 400 also includes a derived object acquisition module 420, configured to acquire a derived object for the knowledge graph in response to at least one second object identifier among a plurality of object identifiers not matching a plurality of base object identifiers. The derived object is determined based on the aggregation degree of the target identifier cluster corresponding to at least one second object identifier. The device 400 also includes an association determination module 430, configured to determine the association between the derived object and the base object corresponding to the base object identifier. The association includes the level of the derived object in the knowledge graph. The device 400 also includes a sorting information providing module 440, configured to provide sorting information indicating the order of multiple objects associated with multiple object identifiers, the order being determined at least based on the hierarchy of the multiple objects in the knowledge graph and the corresponding multi-dimensional evaluation data of the multiple objects, the multiple objects including at least a first object and derived objects.

[0123] The modules included in device 400 can be implemented in various ways, including software, hardware, firmware, or any combination thereof. In some embodiments, one or more units may be implemented using software and / or firmware, such as machine-executable instructions stored on a storage medium. In addition to or as an alternative to machine-executable instructions, some or all of the units in device 400 may be implemented at least partially by one or more hardware logic components. By way of example and not limitation, exemplary types of hardware logic components that may be used include field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-chips (SoCs), complex programmable logic devices (CPLDs), and so on.

[0124] Figure 5 A block diagram of an electronic device 500 in which one or more embodiments of the present disclosure may be implemented is shown. It should be understood that... Figure 5 The electronic device 500 shown is merely exemplary and should not be construed as limiting the functionality and scope of the embodiments described herein. Figure 5 The electronic device 500 shown can be used to achieve Figure 1 Electronic devices 110 or Figure 5 The device 500.

[0125] like Figure 5As shown, electronic device 500 is in the form of a general-purpose computing device. Components of electronic device 500 may include, but are not limited to, one or more processing units or processors 510, memory 520, storage devices 530, one or more communication units 540, one or more input devices 550, and one or more output devices 560. Processor 510 may be a physical or virtual processor and is capable of performing various processes according to programs stored in memory 520. In a multiprocessor system, multiple processing units execute computer-executable instructions in parallel to improve the parallel processing capability of electronic device 500.

[0126] Electronic device 500 typically includes multiple computer storage media. Such media can be any accessible media that is accessible to electronic device 500, including but not limited to volatile and non-volatile media, removable and non-removable media. Memory 520 can be volatile memory (e.g., registers, cache, random access memory (RAM)), non-volatile memory (e.g., read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory), or some combination thereof. Storage device 530 can be removable or non-removable media and can include machine-readable media, such as flash drives, disks, or any other media that can be used to store information and / or data and can be accessed within electronic device 500.

[0127] Electronic device 500 may further include additional removable / non-removable, volatile / non-volatile storage media. Although not explicitly stated... Figure 5 As shown, disk drives for reading from or writing to removable, non-volatile disks (e.g., "floppy disks") and optical disk drives for reading from or writing to removable, non-volatile optical disks can be provided. In these cases, each drive can be connected to a bus (not shown) via one or more data media interfaces. Memory 520 may include computer program product 525 having one or more program modules configured to perform various methods or actions of various embodiments of this disclosure.

[0128] The communication unit 540 enables communication with other computing devices via a communication medium. Additionally, the functionality of the components of the electronic device 500 can be implemented as a single computing cluster or multiple computing machines capable of communicating via communication connections. Therefore, the electronic device 500 can operate in a networked environment using logical connections to one or more other servers, network personal computers (PCs), or another network node.

[0129] Input device 550 can be one or more input devices, such as a mouse, keyboard, trackball, etc. Output device 560 can be one or more output devices, such as a monitor, speaker, printer, etc. Electronic device 500 can also communicate with one or more external devices (not shown) via communication unit 540 as needed. These external devices include storage devices, display devices, etc., and can communicate with one or more devices that enable user interaction with electronic device 500, or with any device that enables electronic device 500 to communicate with one or more other computing devices (e.g., network card, modem, etc.). Such communication can be performed via input / output (I / O) interfaces (not shown).

[0130] According to an exemplary implementation of this disclosure, a computer-readable storage medium is provided that stores computer-executable instructions thereon, wherein the computer-executable instructions are executed by a processor to implement the methods described above. According to an exemplary implementation of this disclosure, a computer program product is also provided, which is tangibly stored on a non-transitory computer-readable medium and includes computer-executable instructions, which are executed by a processor to implement the methods described above.

[0131] Various aspects of this disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatuses, devices, and computer program products implemented according to this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0132] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processing unit of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0133] Computer-readable program instructions can be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions that execute on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0134] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0135] Various implementations of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed implementations. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described implementations. The terminology used herein is chosen to best explain the principles, practical applications, or improvements to technology in the market, or to enable others skilled in the art to understand the various implementations disclosed herein.

Claims

1. A method for data processing, comprising: In response to a first object identifier among multiple object identifiers matching a baseline object identifier in a knowledge graph, a correspondence between the first object identifier and the baseline object identifier is determined, the correspondence including the level of the first object corresponding to the first object identifier in the knowledge graph; In response to a mismatch between at least one second object identifier among the plurality of object identifiers and the plurality of baseline object identifiers, a derived object for the knowledge graph is obtained, the derived object being determined based on the aggregation degree of the target identifier cluster corresponding to the at least one second object identifier; Determine the association between the derived object and the base object corresponding to the base object identifier, wherein the association includes the hierarchy of the derived object in the knowledge graph; as well as The system provides sorting information indicating the order of multiple objects associated with multiple object identifiers. The order is determined at least based on the hierarchy of the multiple objects in the knowledge graph and the corresponding multi-dimensional evaluation data of the multiple objects. The multiple objects include at least the first object and the derived object.

2. The method of claim 1, wherein the first object identifier is determined to match the reference object identifier by: In response to the similarity between the first object identifier and the reference object identifier satisfying a similarity threshold, it is determined that the first object identifier matches the reference object identifier, the similarity threshold being determined based on the category to which the reference object identifier belongs.

3. The method of claim 2, wherein the similarity is determined based on at least one of semantic similarity, feature similarity, and text similarity between the first object identifier and the reference object identifier.

4. The method of claim 1, wherein the plurality of object identifiers are obtained based on preprocessing of a plurality of initial object identifiers, the preprocessing comprising at least one of the following: Remove redundant characters from each initial object identifier; and Based on a pre-defined invalid semantic lexicon, remove non-entity semantic characters from each initial object identifier.

5. The method according to claim 1, wherein obtaining derived objects for the knowledge graph comprises: The target identifier cluster is determined by performing clustering processing on the at least one second object identifier; The aggregation degree is determined based at least on the information entropy of the target identifier cluster and the number of at least one e-commerce store, wherein the number indicates the number of e-commerce stores including the object corresponding to the target identifier cluster; as well as Based on the aggregation degree and the aggregation degree threshold range, the derived object for the knowledge graph is obtained.

6. The method according to claim 5, wherein the aggregation degree threshold range includes a first aggregation degree threshold and a second aggregation degree threshold, the first aggregation degree threshold being greater than the second aggregation degree threshold, and obtaining the derived object for the knowledge graph includes: In response to the aggregation degree being greater than the first aggregation degree threshold, the derived object is determined as a first-level derived object, and the first-level derived object belongs to the same level as the base object; In response to the aggregation degree being greater than the second aggregation degree threshold and less than the first aggregation degree threshold, the derived object is determined as a second-level derived object, and the second-level derived object is a child of the base object; or In response to the aggregation degree being less than a second aggregation degree threshold, the derived object is determined as a predetermined derived object, which is associated with the attributes of the e-commerce store.

7. The method according to claim 6, wherein the attributes of the second-level derived object are determined based on the attributes of the parent object corresponding to the second-level derived object in the knowledge graph.

8. The method according to claim 1, wherein the level corresponding to the base object is higher than the level corresponding to the derived object.

9. The method according to claim 1, wherein the plurality of object identifiers include a plurality of dish names, the plurality of objects include a plurality of dishes corresponding to the plurality of dish names, and the reference object identifier is determined based on a recipe standard file.

10. The method of claim 1, wherein the attributes of the plurality of objects are determined based on attribute information of the e-commerce stores associated with the plurality of object identifiers; and the method further comprises: In response to updates to the information or review data of the e-commerce store, update the attribute of at least one of the plurality of objects.

11. An apparatus for data processing, comprising: The correspondence determination module is configured to determine the correspondence between the first object identifier and the base object identifier in the knowledge graph in response to a first object identifier among a plurality of object identifiers being matched with a base object identifier in the knowledge graph. The correspondence includes the level of the first object corresponding to the first object identifier in the knowledge graph. The derived object acquisition module is configured to acquire a derived object for the knowledge graph in response to at least one second object identifier among the plurality of object identifiers not matching the plurality of baseline object identifiers, wherein the derived object is determined based on the aggregation degree of the target identifier cluster corresponding to the at least one second object identifier; The association determination module is configured to determine the association between the derived object and the base object corresponding to the base object identifier, wherein the association includes the hierarchy of the derived object in the knowledge graph; as well as A sorting information providing module is configured to provide sorting information indicating the order of multiple objects associated with multiple object identifiers, the order being determined at least based on the hierarchy of the multiple objects in the knowledge graph and corresponding multi-dimensional evaluation data of the multiple objects, the multiple objects including at least the first object and the derived object.

12. An electronic device, comprising: At least one processor; as well as At least one memory coupled to the at least one processor and storing instructions for execution by the at least one processor, the instructions causing the electronic device to perform the method according to any one of claims 1 to 10 when executed by the at least one processor.

13. A computer-readable storage medium having a computer program stored thereon, the computer program being executable by a processor to implement the method according to any one of claims 1 to 10.

14. A computer program product tangibly stored in a computer storage medium and comprising computer-executable instructions that, when executed by a device, cause the device to perform the method according to any one of claims 1 to 10.

Images