Object identification method and device, computer storage medium, and electronic device

Abstract

The present disclosure relates to the field of data processing technology, and provides an object identification method, an object identification device, a computer storage medium, and an electronic device, wherein the object identification method includes: roughly matching the objects according to the acquired object attribute information to determine the rough matching pairs of the objects; eliminating the wrong matching pairs in the rough matching pairs to obtain candidate matching pairs; determining the associated objects corresponding to the candidate objects according to the similarity of the objects in the candidate matching pairs; constructing a first feature vector corresponding to the candidate objects, and constructing a second feature vector corresponding to the associated objects; if the vector distance between the first feature vector and the second feature vector is less than a first preset threshold, it is determined that the candidate object is similar to the associated object. The object identification method in the present disclosure can solve the technical problem that only the feature information between two similar objects is considered in the prior art, while ignoring the similar objects corresponding to the two commodities, thereby ensuring the accuracy of identification.

Description

Background Art

[0002] With the rapid development of computer and Internet technology, related e-commerce platforms are also booming. For an e-commerce platform, if it can accurately understand the differences and similarities between its own products and those of its competitors, and carry out targeted pricing and promotion based on the differences in the products, it will undoubtedly ensure the healthy and positive operation of the e-commerce platform.

[0003] At present, the similarity between two products is generally obtained. If the similarity is greater than a preset threshold, the products are considered similar. However, in actual application scenarios, there may be a situation where the similarity between product A and product B is greater than the preset threshold, the similarity between product B and product C is greater than the preset threshold, and the similarity between product A and product C is less than the preset threshold. It can be seen that the processing method in the prior art does not take into account the processing of relatively similar products.

[0004] In view of this, there is an urgent need in the art to develop a new method and device for identifying objects.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure. Summary of the invention

[0006] The purpose of the present disclosure is to provide an object identification method, an object identification device, a computer storage medium and an electronic device, thereby avoiding the defect of low identification accuracy in the prior art at least to a certain extent.

[0007] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0008] According to a first aspect of the present disclosure, a method for identifying an object is provided, comprising: roughly matching the objects according to acquired attribute information of the objects to determine a rough matching pair of the objects; eliminating mismatching pairs in the rough matching pairs to obtain candidate matching pairs; determining associated objects corresponding to the candidate objects according to the similarity of the objects in the candidate matching pairs; constructing a first feature vector corresponding to the candidate objects, and constructing a second feature vector corresponding to the associated objects; if the vector distance between the first feature vector and the second feature vector is less than a first preset threshold, determining that the candidate object is similar to the associated object.

[0009] In an exemplary embodiment of the present disclosure, the rough matching of items based on the acquired item attribute information to determine the rough matching pair of the items includes: filtering invalid information contained in the item attribute information; after the filtering, acquiring attribute name information and attribute value information contained in the item attribute information; and rough matching of the items based on the attribute name information and the attribute value information to determine the rough matching pair.

[0010] In an exemplary embodiment of the present disclosure, determining the associated items corresponding to the candidate items based on the similarity of the items in the candidate matching pairs includes: if the similarity of the items in the candidate matching pairs is greater than a second preset threshold, constructing an association relationship map of the items; and determining the associated items corresponding to the candidate items based on the association relationship map.

[0011] In an exemplary embodiment of the present disclosure, the method further includes: if the object similarity is less than the second preset threshold, eliminating the candidate matching pair.

[0012] In an exemplary embodiment of the present disclosure, the method further includes: obtaining a set of similar items corresponding to each of the associated items; obtaining a correlation coefficient between the sets of similar items; and eliminating the associated items whose correlation coefficient is less than a third preset threshold.

[0013] In an exemplary embodiment of the present disclosure, the method further includes: taking the candidate item and the associated item as target objects; randomly acquiring a target matching pair in the target object; and if the mismatching pair exists in the target matching pair, removing the associated item in the mismatching pair.

[0014] In an exemplary embodiment of the present disclosure, the method further includes: identifying the acquired object tag information according to an entity recognition algorithm to determine the object attribute information.

[0015] According to a second aspect of the present disclosure, there is provided an object identification device, comprising: a coarse matching module, used to perform coarse matching on the objects according to acquired object attribute information to determine coarse matching pairs of the objects; a rejection module, used to reject mismatched pairs in the coarse matching pairs to obtain candidate matching pairs; a candidate object determination module, used to determine associated objects corresponding to the candidate objects according to the similarity of the objects in the candidate matching pairs; a vector construction module, used to construct a first feature vector corresponding to the candidate objects, and to construct a second feature vector corresponding to the associated objects; and a similar object determination module, used to determine that the candidate objects are similar to the associated objects if the vector distance between the first feature vector and the second feature vector is less than a first preset threshold.

[0016] According to a third aspect of the present disclosure, there is provided a computer storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for identifying an object described in the first aspect is implemented.

[0017] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the object identification method described in the first aspect above by executing the executable instructions.

[0018] It can be seen from the above technical solutions that the object identification method, object identification device, computer storage medium and electronic device in the exemplary embodiments of the present disclosure have at least the following advantages and positive effects:

[0019] In the technical solutions provided by some embodiments of the present disclosure, on the one hand, rough matching of items is performed according to the acquired item attribute information to determine the rough matching pairs of items, and the mismatched pairs in the above rough matching pairs are eliminated to obtain candidate matching pairs, which can solve the technical problem that the clustering result data cluster formed by relying only on the clustering features pre-defined in the algorithm for cluster matching in the prior art is too large and has poor flexibility. Further, according to the similarity of the items in the candidate matching pairs, the associated items corresponding to the candidate items are determined, which can solve the technical problem that the prior art only considers the feature information between two similar items and ignores the similar items corresponding to the two commodities, resulting in insufficient recognition accuracy, and expands the coverage of the reference items selected during recognition to ensure recognition accuracy. On the other hand, the first feature vector corresponding to the candidate item is constructed, and the second feature vector corresponding to the associated item is constructed. If the vector distance between the first feature vector and the second feature vector is less than the first preset threshold, it is determined that the candidate item is similar to the associated item, and the features of the item can be digitized, and the recognition results obtained in the previous step can be simplified to further ensure recognition accuracy.

[0020] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0022] Figure 1 A schematic diagram showing a flow chart of a method for identifying an object in an exemplary embodiment of the present disclosure;

[0023] Figure 2 A schematic diagram showing a flow chart of a method for identifying an object in another exemplary embodiment of the present disclosure;

[0024] Figure 3 A schematic diagram showing a flow chart of a method for identifying an object in another exemplary embodiment of the present disclosure;

[0025] Figure 4 A schematic diagram showing the overall process of a method for identifying an object in an exemplary embodiment of the present disclosure;

[0026] Figure 5 A schematic diagram showing the structure of an object identification device in an exemplary embodiment of the present disclosure;

[0027] Figure 6 A schematic diagram showing the structure of a computer storage medium in an exemplary embodiment of the present disclosure;

[0028] Figure 7 A schematic structural diagram of an electronic device in an exemplary embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0029] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0030] The terms "a", "an", "the" and "said" are used in this specification to indicate the presence of one or more elements / components / etc.; the terms "including" and "having" are used to express an open-ended inclusion and mean that additional elements / components / etc. may exist in addition to the listed elements / components / etc.; the terms "first" and "second" etc. are used only as labels and are not intended to limit the quantity of their objects.

[0031] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and their repeated descriptions will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities.

[0032] At present, the similarity between two products is generally obtained. If the similarity is greater than a preset threshold, the products are considered similar. However, in actual application scenarios, there may be a situation where the similarity between product A and product B is greater than the preset threshold, the similarity between product B and product C is greater than the preset threshold, and the similarity between product A and product C is less than the preset threshold. It can be seen that the processing method in the prior art does not take into account the processing of relatively similar products.

[0033] In the embodiments of the present disclosure, firstly, a method for identifying an object is provided, which at least to some extent overcomes the defect of low identification accuracy of the method for identifying an object provided in the prior art.

[0034] Figure 1 A schematic flow chart of an object identification method in an exemplary embodiment of the present disclosure is shown. The object identification method may be executed by a server that identifies similar objects.

[0035] refer to Figure 1 According to an embodiment of the present invention, a method for identifying an object includes the following steps:

[0036] Step S110, roughly matching the objects according to the acquired object attribute information to determine a roughly matching pair of the objects;

[0037] Step S120, eliminating the wrong matching pairs in the rough matching pairs to obtain candidate matching pairs;

[0038] Step S130, determining the associated items corresponding to the candidate items according to the similarities of the items in the candidate matching pairs;

[0039] Step S140, constructing a first feature vector corresponding to the candidate item, and constructing a second feature vector corresponding to the associated item;

[0040] Step S150: If the vector distance between the first feature vector and the second feature vector is less than a first preset threshold, it is determined that the candidate item is similar to the associated item.

[0041] exist Figure 1In the technical solution provided by the illustrated embodiment, on the one hand, the items are roughly matched according to the acquired item attribute information to determine the rough matching pairs of the items, and the mismatched pairs in the above rough matching pairs are eliminated to obtain candidate matching pairs, which can solve the technical problem that the clustering result data cluster formed by relying only on the clustering features pre-defined in the algorithm for cluster matching in the prior art is too large and has poor flexibility. Further, according to the item similarity in the candidate matching pair, the associated items corresponding to the candidate items are determined, which can solve the technical problem that the prior art only considers the feature information between two similar items and ignores the similar items corresponding to the two commodities, resulting in insufficient recognition accuracy, and expands the coverage of the reference items selected during recognition to ensure recognition accuracy. On the other hand, the first feature vector corresponding to the candidate item is constructed, and the second feature vector corresponding to the associated item is constructed. If the vector distance between the first feature vector and the second feature vector is less than the first preset threshold, it is determined that the candidate item is similar to the associated item, and the features of the item can be digitized, and the recognition results obtained in the previous step can be simplified to further ensure recognition accuracy.

[0042] The following Figure 1 The specific implementation process of each step is described in detail:

[0043] In an exemplary embodiment of the present disclosure, the above-mentioned items may be items on sale by e-commerce platform A (considered as benchmark items for the sake of distinction), or items on sale by a competitor of e-commerce platform A (e-commerce platform B) (considered as competing items for the sake of distinction).

[0044] In an exemplary embodiment of the present disclosure, the item label information (e.g., text, pictures, etc. printed on the outer packaging of the item) may be obtained first, and the item label information may be identified based on an entity recognition algorithm to determine the item attribute information (i.e., the characteristics corresponding to the properties of the item, e.g., first-line brands, powders, cosmetics, liquids, etc. are all item attribute information). Exemplarily, the entity recognition algorithm may be a regular expression and a named entity recognition algorithm.

[0045] In an exemplary embodiment of the present disclosure, specifically, regular expressions (Regular Expression, also known as regular expressions, often abbreviated as regex, regexp or RE in the code, which can be used to retrieve and replace text that conforms to the rules) can be used to identify entity types with obvious rules such as time and currency name contained in the item label information to obtain the first category of item attribute information (production date, sales price, shelf life, etc.).

[0046] In an exemplary embodiment of the present disclosure, specifically, based on a named entity recognition algorithm (Named Entity Recognition, referred to as: NER, refers to the identification of named reference items from text, such as: names of people, places, names of organizations, etc.), entity types with unclear rules contained in the item tag information can be identified to obtain a preliminary recognition result. In addition, based on the above-mentioned named entity recognition algorithm, unstructured information in the item tag information can be mined, such as: texts of all formats, pictures, XML (eXtensible Markup Language, referred to as: XML, extensible markup language), HTML (HyperText Markup Language, referred to as: HTML, hypertext markup language), images and audio / video information, etc., and the effective information contained in the unstructured information is supplemented to the above-mentioned preliminary recognition result as the second type of item attribute information (such as: item type information, item usage information and item ingredient information) finally obtained.

[0047] In an exemplary embodiment of the present disclosure, the first category of item attribute information and the second category of item attribute information may be combined as the item attribute information. Exemplarily, the item attribute information finally obtained may include the title of the item (such as Gree air conditioner, Midea air conditioner), sales information (price), extended information (such as manufacturer, warranty period, shelf life, etc.), specification information (such as length, width, height), sales pictures, etc.

[0048] Continue to refer Figure 1 In step S110, the objects are roughly matched according to the acquired object attribute information to determine a rough matching pair of the objects.

[0049] In an exemplary embodiment of the present disclosure, after acquiring the above-mentioned item attribute information, rough matching processing can be performed on the items based on the item attribute information and multiple preset knowledge databases to determine a rough matching pair between the above-mentioned benchmark product and the competing product.

[0050] In an exemplary embodiment of the present disclosure, a plurality of knowledge databases may be pre-constructed, for example, an invalid word library, a replacement rule library, a synonym word library, a hyponym word library, and a mutually exclusive word library.

[0051] In an exemplary embodiment of the present disclosure, a large amount of item tag information (including sales attributes, extended attributes, specification attributes, and image attributes) can be obtained from the internal database of each existing e-commerce platform, and then the item attribute information (including attribute name information and attribute value information) contained in the above item tag information can be extracted, wherein the attribute name information can be various item types, such as: mobile phone, computer, stereo, etc. The attribute value information can be the item content contained under the attribute name information, such as: Apple mobile phone, Huawei mobile phone, etc.

[0052] In an exemplary embodiment of the present disclosure, different e-commerce platforms may set different names for substantially the same attributes. For example, the attribute name "brand" set by e-commerce platform A may be represented as "Brand name" on e-commerce platform B. Therefore, the extracted attribute name information with different names but the same substantial content may be normalized. For example, the attribute name information "brand" and the attribute name information "Brand name" may be processed as the same attribute name information "brand". Furthermore, all attribute name information (for example: "place of origin, sales price") in e-commerce platform A and e-commerce platform B may be extracted, and all attribute value information corresponding to all attribute name information may be extracted and stored together accordingly.

[0053] In an exemplary embodiment of the present disclosure, in view of the limitations of items within the e-commerce platform, some attribute value information may also be obtained from an external knowledge base (external resources such as Baidu and industry knowledge graphs), for example: attribute name information and attribute value information that do not exist in both e-commerce platforms A and B. Furthermore, the attribute name information obtained from the external knowledge base may be merged with the attribute name information obtained from the internal database, and the attribute value information obtained from the external knowledge base may be merged with the attribute value information obtained from the internal knowledge base, to ensure that the acquired data is more complete.

[0054] In an exemplary embodiment of the present disclosure, after obtaining complete attribute name information and attribute value information, a replacement rule can be constructed for all attribute value information, for example, the attribute value information "memory" and "storage component" can be replaced. Composite attribute value information can also be split, for example, "mobile phone memory" can be split into "mobile phone" and "memory" to ensure the fine granularity of the data.

[0055] In an exemplary embodiment of the present disclosure, after the acquired data is processed as above, the above database (invalid word library, replacement rule library, synonym word library, hyponym word library and mutually exclusive word library) can be constructed based on the processed data.

[0056] In an exemplary embodiment of the present disclosure, the invalid word library can be used to filter and remove invalid words in the item attribute information. Invalid information is words that will not affect the main properties of the item after being removed, such as "other, most, most cases".

[0057] In an exemplary embodiment of the present disclosure, the replacement rule library can be used to determine whether the acquired item attribute information is of the same granularity, that is, whether it is at the same level of detail. For example, referring to the relevant explanations of the above steps, when the item attribute information "mobile network type" of e-commerce platform A is "all network 4G (the fourth generation mobile phone mobile communication standard, abbreviation: 4G)", and the item attribute information "mobile network type" of e-commerce platform B is "China Mobile 4G, China Unicom 4G, China Telecom 4G", the replacement rule library can determine that the above item attribute information belongs to the same granularity.

[0058] In an exemplary embodiment of the present disclosure, a synonym library can be used to determine whether item attribute information is the same or whether there is a subordinate relationship. For example, when the item attribute information "material" of e-commerce platform A is "polyester fiber", and the item attribute information "material" of e-commerce platform C is "polyester", the synonym library can determine that the above item attribute information is the same.

[0059] In an exemplary embodiment of the present disclosure, a hyponymous and hyponymous vocabulary is used to determine whether item attribute information is of the same level. For example, when the item attribute information of e-commerce platform A is "home appliances" and the item attribute information of e-commerce platform B is "air conditioner", it can be determined that the above item attribute information belongs to different word levels.

[0060] In an exemplary embodiment of the present disclosure, a mutually exclusive vocabulary is used to determine whether the above-mentioned item attribute information belongs to the same type of classification. For example, when the "user group" of the item attribute information of e-commerce platform A is "middle-aged people" and the "user group" of the item attribute information of e-commerce platform B is "infants and young children", it can be determined that the above-mentioned item attribute information belongs to different types of classification.

[0061] In an exemplary embodiment of the present disclosure, after obtaining the attribute information of the item, the items can be roughly matched based on the above database to determine the rough matching pairs of the items. Figure 2 , Figure 2 A schematic diagram of a method for identifying an object in an exemplary embodiment of the present disclosure is shown, specifically showing a schematic diagram of a method for roughly matching objects according to object attribute information to determine a rough matching pair of objects. Figure 2 Step S110 is explained.

[0062] In step S201, invalid information contained in the item attribute information is filtered.

[0063] In an exemplary embodiment of the present disclosure, after obtaining the item attribute information, the invalid information contained in the item attribute information can be filtered based on the invalid word library. Specifically, the invalid word library can delete the detected invalid information to reduce the amount of data to be processed, increase the data processing speed, and reduce the memory usage of the system.

[0064] In step S202, after the filtering process, the attribute name information and the attribute value information included in the item attribute information are obtained.

[0065] In an exemplary embodiment of the present disclosure, after filtering the item attribute information, the attribute name information and attribute value information contained in the item attribute information can be obtained. For example, reference can be made to Table 1, which shows the schematic content of the attribute name information and the corresponding attribute value information obtained in the present disclosure.

[0066]

[0067] Table 1

[0068] Among them, the above-mentioned Apple mobile phone A1, Huawei mobile phone A2 and Apple computer A3 are the products on sale on e-commerce platform A. The above-mentioned Huawei mobile phones B1, B2, B3, B4, B4 and Huawei computer B6 are the products on sale on e-commerce platform B.

[0069] In step S203, the objects are roughly matched according to the attribute name information and the attribute value information to determine the roughly matched pair.

[0070] In an exemplary embodiment of the present disclosure, after obtaining the above-mentioned attribute name information and attribute value information, the above-mentioned items can be roughly matched based on the above-mentioned invalid word library, replacement rule library, synonym word library, upper and lower relationship word library and mutually exclusive word library according to the above-mentioned attribute name information and attribute value information to determine a rough matching pair. Exemplarily, the items with the attribute name information of "mobile phone" can be matched into one category, and the items with the attribute name information of "computer" can be matched into another category. Specifically, taking the matching of items with the attribute name information of "mobile phone" as an example for explanation, and continuing to refer to Table 1, the obtained rough matching pairs can be: S1 (Apple mobile phone A1-Huawei mobile phone B1), S2 (Huawei mobile phone A2-Huawei mobile phone B2, Huawei mobile phone B3, Huawei mobile phone B4, Huawei mobile phone B5).

[0071] Continue to refer Figure 1 In step S120, the incorrect matching pairs in the rough matching pairs are eliminated to obtain candidate matching pairs.

[0072] In an exemplary embodiment of the present disclosure, after obtaining the above-mentioned rough matching pairs, the mismatching pairs in the above-mentioned rough matching pairs can be eliminated to obtain candidate matching pairs. Exemplarily, referring to the relevant explanation of the above-mentioned step S120, the items contained in the above-mentioned rough matching pair S1 are Apple mobile phone A1 and Huawei mobile phone B1. Although the two items come from different e-commerce platforms, their mobile phone manufacturers are different. Therefore, there must be essential differences and they cannot be similar items. Therefore, the "Apple mobile phone A1-Huawei mobile phone B1" in the above-mentioned rough matching pair S1 can be regarded as a mismatching pair and eliminated, and the rough matching pair S2 (Huawei mobile phone A2-Huawei mobile phone B2, Huawei mobile phone B3, Huawei mobile phone B4, Huawei mobile phone B5) can be used as a candidate matching pair.

[0073] In an exemplary embodiment of the present disclosure, the above-mentioned candidate matches can be further split into candidate matching pair 1 (Huawei mobile phone A2-Huawei mobile phone B2), candidate matching pair 2 (Huawei mobile phone A2-Huawei mobile phone B3), candidate matching pair 3 (Huawei mobile phone A2-Huawei mobile phone B4) and candidate matching pair 4 (Huawei mobile phone A2-Huawei mobile phone B5).

[0074] In the exemplary embodiment of the present disclosure, by eliminating the above-mentioned mismatched pairs, it is possible to eliminate the situation where the attributes of a certain dimension are the same but the essence of the items are different, thereby solving the technical problem in the prior art that cluster matching only relies on the clustering features pre-defined in the algorithm, resulting in the formation of clustering result data clusters that are too large and have poor flexibility, thereby improving the matching accuracy.

[0075] In step S130, the associated items corresponding to the candidate items are determined according to the similarities of the items in the candidate matching pairs.

[0076] In an exemplary embodiment of the present disclosure, after determining the candidate matching pair, the item similarity in the candidate matching pair can be obtained, and the associated items corresponding to the candidate items can be determined according to the numerical value of the item similarity. Thus, the technical problem that only the feature information between two similar items is considered in the prior art, while the similar items corresponding to the two commodities are ignored, resulting in insufficient recognition accuracy can be solved, and the coverage of the reference items selected during recognition can be expanded to ensure recognition accuracy.

[0077] In an exemplary embodiment of the present disclosure, specifically, the item similarities in the above-mentioned candidate matching pairs can be calculated. Specifically, the item similarities between Huawei mobile phone A2 and Huawei mobile phone B2 in the above-mentioned candidate matching pair 1 can be calculated (for example: 92%), and the item similarities between Huawei mobile phone A2 and Huawei mobile phone B3 in the above-mentioned candidate matching pair 2 can be calculated (for example: 91%), the item similarities between Huawei mobile phone A2 and Huawei mobile phone B4 in the above-mentioned candidate matching pair 3 can be calculated (for example: 90%), and the item similarities between Huawei mobile phone A2 and Huawei mobile phone B5 in the above-mentioned candidate matching pair 4 can be calculated (for example: 83%).

[0078] In an exemplary embodiment of the present disclosure, a second preset threshold (e.g., 85%) may be set, and if the similarity of the items is less than the second preset threshold, the candidate matching pair may be eliminated. If 83% is less than 85%, the candidate matching pair 4 may be eliminated.

[0079] In an exemplary embodiment of the present disclosure, if the similarity of the above items is greater than the second preset threshold, a graph of association relationships between the items can be constructed. For example, the similarities of the above items can be sorted from large to small according to the values. Then, first connect Huawei mobile phone A2 with item B2 with the largest similarity value, then connect Huawei mobile phone B2 with item B3 with the second largest similarity value, and then connect Huawei mobile phone B3 with item B4 with the third largest similarity value, and then connect all items with similarities greater than the second preset threshold in sequence.

[0080] In an exemplary embodiment of the present disclosure, similarly, the above-mentioned items (Huawei mobile phones B2, B3, B4) also have corresponding multiple matching items (with a similarity greater than a second preset threshold). Similarly, multiple matching items corresponding to the items (Huawei mobile phones B2, B3, B4) can also be obtained and connected in sequence to form the above-mentioned association relationship map.

[0081] In an exemplary embodiment of the present disclosure, after the above association relationship map is constructed, any item can be selected as a candidate item, and then the items directly connected to the above candidate item are determined as its associated items. For example, taking the determined candidate item S as an example, the determined associated items can be S1, S2, S3 and S4.

[0082] In an exemplary embodiment of the present disclosure, after determining the above-mentioned associated items, the above-mentioned candidate items and their associated items can be used as target objects, and then, a target matching pair in the target object can be randomly obtained, that is, any two items in the target object can be obtained as target matching pairs, and it is determined whether there is a mismatching pair in the target matching pair. Exemplarily, referring to the relevant explanations of the above steps, (S, S1, S2, S3, S4) can be used as the above-mentioned target object, and a target matching pair (S-S1, S-S2, S-S3, S-S4, S1-S2, S1-S3...S3-S4) can be randomly obtained, and then, it is determined whether there is a mismatching pair in the above-mentioned target matching pair, for example: whether they are different items of the same brand, or whether they are the same item for different user groups. Furthermore, if it is determined that the above-mentioned mismatching pair exists, the associated items in the mismatching pair can be eliminated. Exemplarily, if it is determined that the target matching pair S-S4 is different items of the same brand, the associated item S4 can be eliminated, and the associated items corresponding to the candidate item S finally determined include S1, S2, and S3.

[0083] In an exemplary embodiment of the present disclosure, after removing mismatched items from the associated items, a similar item set corresponding to each associated item may be obtained, and then, the associated items may be further streamlined according to the correlation between the similar item sets. For details, please refer to Figure 3 , Figure 3 A flow chart showing a method for identifying an object in another exemplary embodiment of the present disclosure is shown, specifically showing a flow chart showing how to remove related objects based on the correlation between sets of similar objects. Figure 3 Specific implementation methods are explained.

[0084] In step S301, a similar item set corresponding to each of the related items is obtained.

[0085] In an exemplary embodiment of the present disclosure, a set of similar items corresponding to each associated item can be obtained. Specifically, a set of similar items corresponding to each associated item can be determined based on the above-mentioned association relationship graph. For example, the determined set of similar items corresponding to associated item S1 can be set S1{S11, S12, S13...S1n}, the set of similar items corresponding to associated item S2 can be set S2{S21, S22, S23...S2n}, and the set of similar items corresponding to associated item S3 can be set S3{S31, S32, S33...S3n}.

[0086] In step S302, the correlation coefficient between the similar item sets is obtained.

[0087] In an exemplary embodiment of the present disclosure, the correlation coefficient between the above-mentioned combinations of similar items can be obtained. Specifically, the Jaccard correlation coefficient between the above-mentioned sets of similar items can be obtained. Taking the calculation of the Jaccard correlation coefficient of the set S1 and the set S2 as an example, the specific calculation formula can be: The calculation of correlation coefficients between other sets is similar.

[0088] In an exemplary embodiment of the present disclosure, referring to the relevant explanations of the above steps, exemplarily, the correlation coefficient between sets S1 and S2 finally obtained can be J(S1, S2), the correlation coefficient between sets S1 and S3 can be J(S1, S3), and the correlation coefficient between sets S2 and S3 can be J(S2, S3).

[0089] In step S303, the associated items whose correlation coefficient is less than a third preset threshold are eliminated.

[0090] In an exemplary embodiment of the present disclosure, a third preset threshold may be preset, and then, if the above correlation coefficient is less than the third preset threshold, the corresponding associated items may be eliminated. Exemplarily, when the correlation coefficient J(S2, S3) between sets S2 and S3 is less than the third preset threshold, the associated items S2 corresponding to set S2 and the associated items S3 corresponding to set S3 may be eliminated, and then, referring to the relevant explanation of the above step S130, the associated items of the candidate item S finally obtained are S1.

[0091] Continue to refer Figure 1 In step S140, a first feature vector corresponding to the candidate item is constructed, and a second feature vector corresponding to the associated item is constructed.

[0092] In an exemplary embodiment of the present disclosure, referring to the relevant explanation of the above step S303, the determined candidate commodity is commodity S1.

[0093] In an exemplary embodiment of the present disclosure, illustratively, a first feature vector V1 corresponding to the candidate commodity may be constructed based on a neural network algorithm, and a second feature vector V2 corresponding to the associated item may be constructed.

[0094] In step S150, if the vector distance between the first feature vector and the second feature vector is less than a first preset threshold, it is determined that the candidate item is similar to the associated item.

[0095] In an exemplary embodiment of the present disclosure, after constructing the first eigenvector and the second eigenvector, the vector distance (e.g., cosine distance) between the first eigenvector and the second eigenvector may be calculated. Exemplarily, the calculation of the cosine distance between the first eigenvector V1 (1, 3) and the second eigenvector V2 (2, 0) is used as an example for explanation.

[0096] In an exemplary embodiment of the present disclosure, a first preset threshold may be pre-set. For example, the first preset threshold may be set to 0.5. Then, the vector distance may be determined. If it is less than the first preset threshold value 0.5, it can be further determined that the candidate item S is similar to the associated item S1. Thus, the features of the item can be digitized, the recognition result obtained in the previous step can be simplified, and the recognition accuracy can be further ensured.

[0097] In an exemplary embodiment of the present disclosure, when there are multiple second feature vectors, the vector distances between the multiple first feature vectors and the second feature vectors can be calculated, and then the associated items corresponding to the second feature vectors can be sorted in the order of the vector distances from small to large. Furthermore, different numbers of associated items can be selected as similar items of the candidate items according to actual needs.

[0098] In an exemplary embodiment of the present disclosure, Figure 4 The overall flow chart of the object identification method in an exemplary embodiment of the present disclosure is schematically shown. Figure 4 Specific implementation methods are explained.

[0099] In step S401, start;

[0100] In step S402, the item tag information (title, sales attributes, extended attributes, specification attributes, image attributes, etc.) is identified to determine the item attribute information;

[0101] In step S403, according to the above-mentioned item attribute information, the items are roughly matched based on the preset database (stop word library, synonym library, replacement rule library, and upper and lower word library) to determine the rough matching pairs;

[0102] In step S404, it is determined whether there is a mismatched pair in the rough matching pairs;

[0103] In step S405, if there is no mismatched pair, the items in the rough matching pair are determined to be similar items;

[0104] In step S406, if there are any mismatched pairs, the mismatched pairs are eliminated to obtain candidate matching pairs;

[0105] In step S407, the similarity of the items in the candidate matching pairs is calculated, and if the similarity is greater than a second preset threshold, a correlation relationship graph of the items is constructed;

[0106] In step S408, any item is selected as a candidate item, and the associated items corresponding to the candidate item are determined;

[0107] In step S409, the candidate items and the associated items are taken as target objects, target matching pairs in the target objects are randomly obtained, and it is determined whether there are mismatched pairs in the target matching pairs;

[0108] In step S410, if there is, the associated items in the mismatched pair are removed, and the similar item sets corresponding to each associated item are determined;

[0109] In step S411, the Jaccard correlation coefficient corresponding to the similar item set is calculated, and if the correlation coefficient is less than a third preset threshold, the associated items corresponding to the similar item set are removed;

[0110] In step S412, a first feature vector corresponding to the candidate item is constructed, and a second feature vector corresponding to the remaining associated items is constructed;

[0111] In step S413, the cosine distance between the first eigenvector and the second eigenvector is calculated;

[0112] In step S414, the related items are sorted in order of cosine distance from small to large;

[0113] In step S415, the first K related items are determined as similar items corresponding to the candidate items; (the specific value of K can be set according to actual conditions and falls within the scope of protection of the present disclosure)

[0114] In step S416, end.

[0115] The present disclosure also provides an object identification device, Figure 5 A schematic diagram showing the structure of an object identification device in an exemplary embodiment of the present disclosure; Figure 5 As shown, the object recognition device 500 may include a rough matching module 501, a rejection module 502, a candidate object determination module 503, a vector construction module 504 and a similar object determination module 505. Among them:

[0116] The rough matching module 501 is used to perform rough matching on the objects according to the acquired object attribute information to determine rough matching pairs of the objects.

[0117] In an exemplary embodiment of the present disclosure, the coarse matching module is used to identify the acquired object tag information according to the entity recognition algorithm to determine the object attribute information.

[0118] In an exemplary embodiment of the present disclosure, a coarse matching module is used to filter invalid information contained in item attribute information; after filtering, the attribute name information and attribute value information contained in the item attribute information are obtained; and the items are coarsely matched based on the attribute name information and the attribute value information to determine a coarse matching pair.

[0119] The elimination module 502 is used to eliminate the wrong matching pairs in the rough matching pairs to obtain candidate matching pairs.

[0120] In an exemplary embodiment of the present disclosure, the elimination module is used to eliminate the wrong matching pairs in the rough matching pairs to obtain candidate matching pairs.

[0121] The candidate item determination module 503 is used to determine the associated items corresponding to the candidate items according to the similarities of the items in the candidate matching pairs.

[0122] In an exemplary embodiment of the present disclosure, the candidate item determination module is used to construct an item association graph if the item similarity in the candidate matching pair is greater than a second preset threshold; based on the association graph, determine the associated items corresponding to the candidate items.

[0123] In an exemplary embodiment of the present disclosure, the candidate item determination module is configured to eliminate candidate matching pairs if the item similarity is less than a second preset threshold.

[0124] In an exemplary embodiment of the present disclosure, the candidate item determination module is used to obtain a similar item set corresponding to each associated item; obtain a correlation coefficient between the similar item sets; and eliminate associated items whose correlation coefficient is less than a third preset threshold.

[0125] In an exemplary embodiment of the present disclosure, the candidate item determination module is further used to take the candidate item and the associated item as the target object; randomly obtain a target matching pair in the target object; if there is a mismatching pair in the target matching pair, then remove the associated items in the mismatching pair.

[0126] The vector construction module 504 is used to construct a first feature vector corresponding to the candidate item, and to construct a second feature vector corresponding to the associated item.

[0127] In an exemplary embodiment of the present disclosure, the vector construction module is used to construct a first feature vector corresponding to a candidate item, and to construct a second feature vector corresponding to an associated item.

[0128] The similar item determination module 505 is configured to determine that the candidate item is similar to the associated item if the vector distance between the first feature vector and the second feature vector is less than a first preset threshold.

[0129] In an exemplary embodiment of the present disclosure, the similar item determination module is used to determine that the candidate item is similar to the associated item if the vector distance between the first feature vector and the second feature vector is less than a first preset threshold.

[0130] The specific details of each module in the above-mentioned object identification device have been described in detail in the corresponding object identification method, so they will not be repeated here.

[0131] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0132] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0133] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0134] In the exemplary embodiments of the present disclosure, a computer storage medium capable of implementing the above method is also provided. A program product capable of implementing the above method of this specification is stored thereon. In some possible embodiments, various aspects of the present disclosure may also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to perform the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of this specification.

[0135] refer to Figure 6 As shown, a program product 600 for implementing the above method according to an embodiment of the present disclosure is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto, and in this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.

[0136] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0137] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0138] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0139] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0140] In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided.

[0141] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0142] Refer to the following Figure 7 An electronic device 700 according to this embodiment of the present disclosure is described. Figure 7 The electronic device 700 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0143] like Figure 7 As shown, the electronic device 700 is in the form of a general computing device. The components of the electronic device 700 may include but are not limited to: at least one processing unit 710, at least one storage unit 720, and a bus 730 connecting different system components (including the storage unit 720 and the processing unit 710).

[0144] The storage unit stores program codes, which can be executed by the processing unit 710, so that the processing unit 710 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification. For example, the processing unit 710 can perform the following steps: Figure 1As shown in: Step S110, roughly matching the items according to the acquired item attribute information to determine the rough matching pairs of the items; Step S120, eliminating the wrong matching pairs in the rough matching pairs to obtain candidate matching pairs; Step S130, determining the associated items corresponding to the candidate items according to the item similarities in the candidate matching pairs; Step S140, constructing a first feature vector corresponding to the candidate items, and constructing a second feature vector corresponding to the associated items; Step S150, if the vector distance between the first feature vector and the second feature vector is less than a first preset threshold, determining that the candidate item is similar to the associated item.

[0145] The storage unit 720 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 7201 and / or a cache storage unit 7202 , and may further include a read-only storage unit (ROM) 7203 .

[0146] The storage unit 720 may also include a program / utility 7204 having a set (at least one) of program modules 7205, such program modules 7205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0147] Bus 730 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0148] The electronic device 700 may also communicate with one or more external devices 800 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 700, and / or communicate with any device that enables the electronic device 700 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 750. Furthermore, the electronic device 700 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 760. As shown, the network adapter 760 communicates with other modules of the electronic device 700 via a bus 730. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0149] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0150] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present disclosure, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0151] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and examples are to be considered as exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

Claims

1. A method for identifying an object, characterized in that: include: Roughly matching the objects according to the acquired object attribute information to determine a rough matching pair of the objects; Eliminating mismatched pairs in the rough matching pairs to obtain candidate matching pairs; Determining associated items corresponding to the candidate items according to the similarities of the items in the candidate matching pairs; The determining, based on the similarity of the items in the candidate matching pair, the associated items corresponding to the candidate items, comprises: if the similarity of the items in the candidate matching pair is greater than a second preset threshold, constructing an associated relationship graph of the items; and determining, based on the associated relationship graph, the associated items corresponding to the candidate items; Acquire a set of similar items corresponding to each of the associated items; acquire a correlation coefficient between the sets of similar items; and eliminate target associated items whose correlation coefficient is less than a third preset threshold; Constructing a first feature vector corresponding to the candidate item, and constructing a second feature vector corresponding to the associated item; If the vector distance between the first feature vector and the second feature vector is less than a first preset threshold, it is determined that the candidate item is similar to the associated item.

2. The method according to claim 1, characterized in that: The roughly matching the items according to the acquired item attribute information to determine the roughly matching pairs of the items includes: Filtering invalid information contained in the item attribute information; After the filtering process, obtaining the attribute name information and the attribute value information contained in the item attribute information; The objects are roughly matched according to the attribute name information and the attribute value information to determine the rough matching pair.

3. The method according to claim 1, characterized in that The method further comprises: If the object similarity is less than the second preset threshold, the candidate matching pair is eliminated.

4. The method according to claim 1, characterized in that The method further comprises: Taking the candidate items and the associated items as target objects; Randomly obtain a target matching pair from the target object; If the mismatched pair exists in the target matching pair, the associated items in the mismatched pair are removed.

5. The method according to claim 1, characterized in that The method further comprises: The acquired object tag information is identified according to an entity recognition algorithm to determine the object attribute information.

6. An object identification device, characterized in that: include: A rough matching module, used to perform rough matching on the items according to the acquired item attribute information to determine a rough matching pair of the items; A removal module, used for removing the wrong matching pairs in the rough matching pairs to obtain candidate matching pairs; The elimination module is further configured to construct an association relationship map of the items if the similarity of the items in the candidate matching pair is greater than a second preset threshold; and determine the associated items corresponding to the candidate items based on the association relationship map; The elimination module is further used to obtain a set of similar items corresponding to each of the associated items; obtain a correlation coefficient between the sets of similar items; and eliminate target associated items whose correlation coefficient is less than a third preset threshold value; A candidate item determination module, used to determine the associated items corresponding to the candidate items according to the similarity of the items in the candidate matching pairs; A vector construction module, used to construct a first feature vector corresponding to the candidate item, and to construct a second feature vector corresponding to the associated item; The similar item determination module is configured to determine that the candidate item is similar to the associated item if the vector distance between the first feature vector and the second feature vector is less than a first preset threshold.

7. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for identifying an object according to any one of claims 1 to 5 is implemented.

8. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to execute the object identification method according to any one of claims 1 to 5 by executing the executable instructions.

Images