Same product identification method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a same product identification method and device, electronic equipment and a storage medium. The method comprises the steps of determining at least one attribute feature matched with a first article and a second article for the first article and the second article to be subjected to same article identification; for each attribute feature in the at least one attribute feature, determining a value similarity corresponding to the attribute feature according to a first attribute value of the first article under the attribute feature and a second attribute value of the second article under the attribute feature; and according to the value similarity corresponding to the at least one attribute feature, performing same article identification on the first article and the second article. According to the technical scheme provided by the embodiment of the invention, similarity comparison is performed on the attribute values of all the attribute characteristics matched with the first article and the second article, so that even if part of the attribute values are not standard, the standard attribute values can be referred to for identification of the same article, and the accuracy of identification of the same article is improved.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of computer technology, and in particular to a method, device, electronic device and storage medium for identifying same product. Background Art

[0002] With the rapid development and application of the digital economy, physical digitization has gradually been integrated into all aspects of people's lives. For example, people can go online to choose the clothes and food they need without leaving home. After placing an order, the corresponding physical items can be quickly delivered to people through logistics distribution. Therefore, data-based analysis and intelligent scheduling play an increasingly important role in all walks of life.

[0003] At present, many service providers have built a variety of systems for item recommendation based on data analysis, but standardization of items themselves is still a big challenge. It should be noted that the core of standardized management of identical items across platforms or stores is same-item identification, that is, identifying whether two items on different platforms or different stores on the same platform are the same item (i.e., the same product).

[0004] In the process of realizing the present invention, the inventors found that the prior art has the following technical problems: the accuracy of same product identification is not high and needs to be improved. Summary of the invention

[0005] The embodiments of the present invention provide a method, device, electronic device and storage medium for identifying same products, so as to improve the accuracy of identifying same products.

[0006] According to one aspect of the present invention, a method for identifying same product is provided, which may include:

[0007] For a first object and a second object to be identified as the same object, determining at least one attribute feature that matches the first object and the second object;

[0008] For each attribute feature of the at least one attribute feature, according to a first attribute value of the first item under the attribute feature and a second attribute value of the second item under the attribute feature, determine a value similarity corresponding to the attribute feature;

[0009] The first object and the second object are identified as the same object according to the similarity of the values ​​corresponding to the at least one attribute feature.

[0010] According to another aspect of the present invention, a device for identifying same product is provided, which may include:

[0011] An attribute feature determination module, for determining, for a first object and a second object to be identified as identical, at least one attribute feature matching the first object and the second object;

[0012] a value similarity determination module, configured to determine, for each attribute feature of at least one attribute feature, a value similarity corresponding to the attribute feature according to a first attribute value of the first item under the attribute feature and a second attribute value of the second item under the attribute feature;

[0013] The same product identification module is used to identify the first item and the second item as the same item according to the similarity of the values ​​corresponding to at least one attribute feature.

[0014] According to another aspect of the present invention, there is provided an electronic device, which may include:

[0015] at least one processor; and

[0016] a memory communicatively connected to at least one processor; wherein,

[0017] The memory stores a computer program that can be executed by at least one processor. The computer program is executed by at least one processor so that the at least one processor implements the same product identification method provided by any embodiment of the present invention when executing the computer program.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided, on which computer instructions are stored, and the computer instructions are used to enable a processor to implement the same product identification method provided by any embodiment of the present invention when executed.

[0019] The technical solution of the embodiment of the present invention is to identify the first and second items to be identified as the same product by determining at least one attribute feature that matches the first and second items; for each of the at least one attribute feature, according to the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature, determine the value similarity corresponding to the attribute feature; and identify the first and second items as the same product according to the value similarity corresponding to the at least one attribute feature. The above technical solution compares the attribute values ​​of all attribute features that match the first and second items for similarity. In this way, even if some of the attribute values ​​are not standardized, especially the attribute values ​​of key features are not standardized, the same product can be identified by referring to those standardized attribute values, thereby improving the accuracy of the same product identification.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 is a flowchart of a method for identifying same products according to an embodiment of the present invention;

[0023] Figure 2 This is a diagram showing an example of using a pre-trained model in a method for identifying a same product according to an embodiment of the present invention;

[0024] Figure 3 is a flow chart of another method for identifying same products according to an embodiment of the present invention;

[0025] Figure 4 is a flow chart of another method for identifying same products provided according to an embodiment of the present invention;

[0026] Figure 5 is a flow chart of another method for identifying same products provided according to an embodiment of the present invention;

[0027] Figure 6 is a flowchart of an optional example of another method for identifying same product provided according to an embodiment of the present invention;

[0028] Figure 7 is a structural block diagram of a same-product identification device provided according to an embodiment of the present invention;

[0029] Figure 8 It is a schematic diagram of the structure of an electronic device for implementing the same product identification method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. The situations of "target", "original", etc. are similar and will not be repeated here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] Figure 1 This is a flowchart of a method for identifying the same product provided in an embodiment of the present invention. This embodiment can be applied to the case of identifying whether two items are the same product, and is particularly applicable to the case of identifying the same product of items with irregular attribute values. The method can be executed by a device for identifying the same product provided in an embodiment of the present invention, and the device can be implemented by software and / or hardware, and the device can be integrated in an electronic device, and the electronic device can be various user terminals or servers.

[0033] See also Figure 1 The method of the embodiment of the present invention specifically includes the following steps:

[0034] S110: For a first object and a second object to be identified as the same object, determine at least one attribute feature that matches the first object and the second object.

[0035] The first item and the second item can be understood as two items to be identified as the same item, that is, there is a need to identify the same item. In combination with the application scenarios that may be involved in the embodiments of the present invention, the two items can be two commodities in different platforms, different stores on the same platform, or the same store on the same platform.

[0036] The attribute feature can be understood as a feature that matches both items, or that both items have. The number of attribute features can be one, two, or more, and is not specifically limited here. For example, here, the two items are two notebooks, and at least one attribute feature corresponding to them may include brand, memory size, processor model, graphics card model, item size, and appearance color.

[0037] S120. For each attribute feature of the at least one attribute feature, determine the value similarity corresponding to the attribute feature according to the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature.

[0038] Among them, for each attribute feature in at least one attribute feature, the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature are obtained. It should be noted that the essence of the first attribute value and the second attribute value are both attribute values ​​under the attribute feature. Here, the distinction is made in naming only to distinguish the attribute values ​​of different items under the attribute feature, and it is not a specific limitation of their substantive meanings. Exemplarily, continuing with the above example, assuming that the attribute feature is the memory size, the first attribute value can be 4G, and the second attribute value can be 6G. Further, according to the first attribute value and the second attribute value, the value similarity corresponding to the attribute feature is obtained. Exemplarily, the similarity of the first attribute value and the second attribute value can be calculated, so that the obtained calculation result is used as the value similarity corresponding to the attribute feature. It can be understood that the value similarity can be used to characterize the probability that the first attribute value and the second attribute value have the same meaning. For example, "first-level energy efficiency" and "first-level energy consumption" have the same meaning, and their value similarity may be very close to 1.

[0039] It should be noted that this step processes a single attribute feature, that is, after this step, the value similarity corresponding to each attribute feature in at least one attribute feature can be obtained.

[0040] S130: Identify the first item and the second item as the same item based on the similarity of the values ​​corresponding to the at least one attribute feature.

[0041] Among them, when the value similarity corresponding to at least one attribute feature is obtained, the first item and the second item can be identified as the same item based on these value similarities, that is, whether the first item and the second item are the same item can be determined. In practical applications, optionally, the importance corresponding to at least one attribute feature can be obtained, and then the same item can be identified based on the importance and value similarity corresponding to at least one attribute feature; optionally, the item similarity between the first item and the second item can be obtained based on the value similarity corresponding to at least one attribute feature, and then the same item can be identified based on the item similarity; etc., which are not specifically limited here.

[0042] On this basis, further, when the first item and the second item are identified as the same item, the two items can be labeled based on the same code, which is conducive to the standardized management of items. It should be noted that the standardized management of items helps to meet the growing personalized needs of consumers, and can promote the circulation of items and the interconnection of information on a larger scale, thereby exploring and discovering greater business opportunities and value in the process of the circulation of items and the interconnection of information.

[0043] It should be noted that, in actual applications, since the attribute values ​​of items are usually subjectively filled in by store merchants, this may lead to various problems such as redundancy, noise, missing, dislocation, synonyms and spelling errors in these attribute values. The non-standardization of attribute values ​​will seriously affect the accuracy of same product identification. In order to solve the above problems, the embodiment of the present invention proposes a scheme for performing a similarity comparison on the attribute values ​​of all attribute features that match the first item and the second item, that is, it is not a scheme for performing a similarity comparison on the attribute values ​​of the key features in all attribute features. In this way, even if some of these attribute values ​​have non-standard problems, especially the attribute values ​​of the key features have non-standard problems, the same product can be identified by referring to those standardized attribute values, thereby improving the accuracy of same product identification.

[0044] The technical solution of the embodiment of the present invention is to identify the first and second items to be identified as the same product by determining at least one attribute feature that matches the first and second items; for each of the at least one attribute feature, according to the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature, determine the value similarity corresponding to the attribute feature; and identify the first and second items as the same product according to the value similarity corresponding to the at least one attribute feature. The above technical solution compares the attribute values ​​of all attribute features that match the first and second items for similarity. In this way, even if some of the attribute values ​​are not standardized, especially the attribute values ​​of key features are not standardized, the same product can be identified by referring to those standardized attribute values, thereby improving the accuracy of the same product identification.

[0045] An optional technical solution, determining the similarity of values ​​corresponding to the attribute features according to the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature, may include:

[0046] Obtaining a first attribute value of the first item under the attribute feature, and a second attribute value of the second item under the attribute feature;

[0047] Inputting the first attribute value into the pre-trained model to obtain a first vector, and inputting the second attribute value into the pre-trained model to obtain a second vector;

[0048] The vector similarity between the first vector and the second vector is determined, and the vector similarity is used as the value similarity corresponding to the attribute feature.

[0049] Considering that there are no labels for the same product in the same product identification process, synonym discrimination can be performed based on the pre-trained model. Figure 2, the pre-training model can first be pre-trained on a large-scale corpus to learn general knowledge, thereby obtaining knowledge related to language semantics and grammar, and then the model that has learned general knowledge can be transferred to downstream tasks for application or fine-tuning. The knowledge learned in the pre-training stage will be of great help to many downstream tasks. In practical applications, optionally, the above-mentioned pre-training model can be RoBERTa, ALBERT, ERNIE, T5 or Longformer, etc., which are not specifically limited here. It should be noted that the pre-training model can usually provide two functions, encoding (Encode) and decoding (Decode). Based on the pre-training model, any Chinese, English or digital code can be mapped to a vector of fixed length, and the mapped vector can also be decoded and restored to the original value. Exemplarily, the "first-level energy efficiency" can be encoded and output as a fixed-length 0-1 vector, such as a vector of [0,1,1,0…1], through the pre-training model, and [0,1,1,0…1] can also be decoded and restored to "first-level energy efficiency" through the pre-training model. From the above description, it can be seen that when the first attribute value and the second attribute value are obtained, the first attribute value can be input into the pre-trained model to obtain the first vector, and the second attribute value can be input into the pre-trained model to obtain the second vector.

[0050] After encoding the attribute value as a vector, the value similarity between the two attribute values ​​can be further expressed by calculating the vector similarity between the two vectors. In practical applications, optional vector similarity calculation schemes that can be used include Euclidean distance, cosine similarity or cosine distance, etc., which are not specifically limited here. Taking cosine similarity as an example, the calculation formula for cosine similarity cos(θ) between the first vector A and the second vector B is as follows, where n represents the vector dimension:

[0051]

[0052] The above technical solution maps two attribute values ​​into vectors of the same length based on a pre-trained model, and then represents the value similarity corresponding to the attribute features based on the vector similarity between the vectors. The essence of the above vector similarity is semantic similarity, that is, the value similarity is represented by semantic similarity, thereby solving the problem of low accuracy in identifying the same product because the attribute values ​​are synonyms.

[0053] Another optional technical solution is to identify the first item and the second item as the same item according to the similarity of the values ​​corresponding to at least one attribute feature, including:

[0054] Obtaining the importance level corresponding to at least one attribute feature;

[0055] The first object and the second object are identified as the same object according to the value similarity and importance respectively corresponding to the at least one attribute feature.

[0056] Among them, in the process of identifying the same product, considering that the degree of influence (or importance) of each attribute feature in at least one attribute feature on the definition of the same product may be different, here takes the above-mentioned notebook example as an example. Generally, compared with the "appearance color", the "memory size" has a greater degree of influence in the process of identifying the same product. Therefore, the importance corresponding to each attribute feature can be obtained, so as to identify the same product based on the value similarity and importance. Exemplarily, the object similarity between two objects can be obtained based on the value similarity and importance corresponding to at least one attribute feature, and then the same product can be identified based on the object similarity. The above technical scheme further improves the accuracy of identifying the same product by applying the importance corresponding to each attribute feature.

[0057] On this basis, optionally, the same product identification method further includes:

[0058] For each pool item in the item pool where the first item and the second item are located, the acquisition cost of the pool item and the pool attribute value of the pool item under each attribute feature are used as a set of training samples;

[0059] Based on the training samples corresponding to each pool item, the original prediction model is trained to obtain a cost prediction model;

[0060] For each attribute feature of the at least one attribute feature, obtaining feature index data of the attribute feature under a preset feature index obtained during the training process of the cost prediction model;

[0061] According to the characteristic index data respectively corresponding to at least one attribute feature, the importance degree respectively corresponding to at least one attribute feature is obtained.

[0062] In actual applications, considering that there are no same-product labels for all items, and the proportion of items with all correct attribute values ​​is often low, the acquisition cost of the item is used as a label for auxiliary judgment. This is based on the assumption that the same items are not only the same or similar in attribute values, but also have similar acquisition costs. Combined with the application scenarios that may be involved in the embodiments of the present invention, the above acquisition cost can be understood as the transaction price, that is, the selling price of the item.

[0063] In addition, when performing the same product identification, for each item from different platforms, different stores on the same platform, or the same store on the same platform, they can be divided into corresponding item pools according to the categories to which these items belong. The above categories can be notebooks, refrigerators, clothes, or shoes, etc., which are not specifically limited here. The advantage of this setting is that each item belonging to the same category has certain commonalities. By dividing these items belonging to the same category into the same item pool, and then analyzing these items in the same item pool, the importance of each attribute feature corresponding to these items can be obtained.

[0064] On this basis, specifically, for the item pool where the first item and the second item are located, the acquisition cost of each pool item in the item pool and the pool attribute value of the pool item under each attribute feature are obtained. Among them, all pool items in the item pool include at least the first item and the second item. The above-mentioned pool items are the items in the item pool. In order to distinguish them from the first item and the second item, they are named as pool items; the above-mentioned pool attribute values ​​are the attribute values ​​of the pool items under the attribute features. In order to distinguish them from the first attribute value and the second attribute value, they are named as pool attribute values.

[0065] The acquisition cost and the pool attribute values ​​under each attribute feature are used as a set of training samples. After processing each pool item in the item pool, the training samples corresponding to each pool item can be obtained. Further, the original prediction model is trained based on these training samples to obtain a cost prediction model, which is used to predict the acquisition cost of the pool item based on the attribute values ​​of the pool item. In practical applications, the above original prediction model can optionally be a regression model, random forest, LightGBM or other models with prediction functions, which are not specifically limited here.

[0066] Furthermore, for each attribute feature in at least one attribute feature, characteristic indicator data of the attribute feature under a preset characteristic indicator obtained during the training process of the cost prediction model is obtained. The preset characteristic indicator can be understood as a pre-set indicator that can be used to characterize the importance of the attribute feature. In practical applications, it can optionally be the number of times the attribute feature is used, the average information gain, or the shap value, etc., which are not specifically limited here. Further, according to the characteristic indicator data corresponding to at least one attribute feature, their corresponding importance is obtained. The above technical solution evaluates the importance of the attribute feature based on the performance of the attribute feature in the model, thereby achieving the effect of accurately quantifying the importance.

[0067] Figure 3It is a flow chart of another same-product identification method provided in an embodiment of the present invention. This embodiment is optimized based on the above-mentioned technical solutions. In this embodiment, optionally, at least one attribute feature that matches the first item and the second item is determined; the item pool where the first item and the second item are located is determined, and at least one synonymous feature corresponding to the item pool is obtained, wherein the at least one synonymous feature is determined according to the pool feature of each pool item in the item pool; at least one synonymous feature is used as an attribute feature that matches the first item and the second item, respectively, to obtain at least one attribute feature. Among them, the explanations of the terms that are the same or corresponding to the above-mentioned embodiments are not repeated here.

[0068] See also Figure 3 The method of this embodiment may specifically include the following steps:

[0069] S210. For a first item and a second item to be identified as the same item, determine an item pool where the first item and the second item are located, and obtain at least one synonymous feature corresponding to the item pool, wherein the at least one synonymous feature is determined based on a pool feature of each pool item in the item pool.

[0070] Among them, the synonymous feature can be understood as a feature determined according to the pool feature of each pool item in the item pool. It should be noted that the essence of the pool feature is the attribute feature. Here, the distinction is made in the naming in order to distinguish the attribute features of the first item / second item and the attribute features of the pool item, but it is not a specific limitation on its essential meaning; in addition, for each pool item in the item pool, the number of its pool features can be one, two or more, which is not specifically limited here. The number of synonymous features can be one, two or more, and when there are at least two synonymous features, the meanings of each of the at least two synonymous features are different. Obtain at least one synonymous feature corresponding to the item pool.

[0071] S220: Use at least one synonymous feature as an attribute feature matching the first item and the second item, respectively, to obtain at least one attribute feature.

[0072] Among them, at least one synonymous feature is used as an attribute feature matching the first item and the second item, respectively, so as to obtain at least one attribute feature. In other words, the meanings of the attribute features in the at least one attribute feature are different from each other, so that different aspects of information can be obtained by subsequently analyzing the similarity of the values ​​corresponding to each attribute feature, thereby further improving the accuracy of identifying the same product.

[0073] It should be noted that the essence of synonymous features and attribute features is the same, except that the former is relative to the item pool, while the latter is equivalent to the first item and the second item. Therefore, the attribute features corresponding to the first item and the second item also correspond to the item pool. The attribute features corresponding to the item pool mentioned below are the attribute features corresponding to the first item and the second item.

[0074] S230. For each attribute feature of the at least one attribute feature, determine the value similarity corresponding to the attribute feature according to the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature.

[0075] S240: Identify the first item and the second item as the same item based on the similarity of the values ​​corresponding to the at least one attribute feature.

[0076] The technical solution of the embodiment of the present invention is to apply at least one synonymous feature with different meanings corresponding to the item pool as an attribute feature matching the first item and the second item respectively. In this way, by subsequently analyzing the similarity of the values ​​corresponding to each attribute feature, different aspects of information can be obtained, thereby further improving the accuracy of identifying the same item.

[0077] An optional technical solution, the above-mentioned same product identification method further includes:

[0078] For each pool item in the item pool, obtain the feature name of the pool feature of the pool item and the pool attribute value of the pool item under the pool feature;

[0079] For the name features with the same feature name in each pool feature corresponding to the item pool, the value range corresponding to the feature name is obtained according to the values ​​of each pool attribute under the name feature;

[0080] For at least two feature names corresponding to the item pool, the feature names with similar value ranges in the at least two feature names are merged to obtain at least one synonymous name;

[0081] Based on each pool feature, a synonym feature corresponding to at least one synonym name is obtained.

[0082] Among them, for each pool item in the item pool, one or more pool features respectively correspond to these pool features, and these pool features are used as pool features corresponding to the item pool. Exemplarily, assuming that the item pool includes pool items S1, S2, S3 and S4, the pool features of S1 include F1, F2 and F3, wherein the feature name of F1 is N1, the feature name of F2 is N2, and the feature name of F3 is N3; the pool features of S2 include F4, F5 and F6, wherein the feature name of F4 is N1, the feature name of F5 is N2, and the feature name of F6 is N4; the pool features of S3 include F7, F8, F9 and F10, wherein the feature name of F7 is N3, the feature name of F8 is N4, the feature name of F9 is N5, and the feature name of F10 is N6; the pool features of S4 include F11 and F12, wherein the feature name of F11 is N2, and the feature name of F12 is N7, on this basis, the pool features corresponding to the item pool include F1-F12.

[0083] The name feature can be understood as a pool feature with the same feature name among the pool features corresponding to the item pool. Continuing with the above example, for example, F1 and F4 are name features with N1, F2, F5 and F11 are name features with N2, and the situation of N3-N7 is similar, which will not be repeated here. According to the values ​​of each pool attribute under the name feature, the value range corresponding to the feature name is obtained, and the value range can be understood as the range where these pool attribute values ​​are located. For example, still taking the above example, for N1, according to the pool attribute value of S1 under F1 and the pool attribute value of S2 under F4, the value range corresponding to N1 is obtained; for N2, according to the pool attribute value of S1 under F2, the pool attribute value of S2 under F5 and the pool attribute value of S4 under F11, the value range corresponding to N2 is obtained. At this point, the value range corresponding to each feature name can be obtained, such as the value ranges corresponding to N1-N7 in the above example.

[0084] Furthermore, the item pool corresponds to at least two feature names. When the value ranges corresponding to some of the feature names are similar, this indicates that these feature names are synonyms, that is, they have the same meaning but different expressions. At this time, these feature names with similar value ranges can be merged to obtain objection names. In practical applications, the merging process of feature names can be implemented in a variety of ways. For example, the most frequently used high-frequency names among the feature names with similar value ranges are determined, and then the remaining feature names are replaced with high-frequency names. For example, for "energy efficiency level" and "energy consumption level" with similar value ranges, assuming that 100 items use "energy efficiency level" and 20 items use "energy consumption level", the "energy consumption level" of these 20 items can be forcibly replaced with "energy efficiency level"; another example is to merge by constructing a mapping dictionary, that is, mapping the feature names with similar value ranges to high-frequency names. For example, for the above example, a dictionary of {"energy efficiency level": "energy efficiency level", "energy consumption level"} can be constructed, and in subsequent processing, multiple feature names after the colon in the dictionary can be considered to be equivalent to the high-frequency names before the colon; and so on, which are not specifically limited here.

[0085] At this point, after merging all the feature names that can be merged in at least two feature names, at least one objection name can be obtained. Exemplarily, for N1-N7 in the above example, assuming that the value ranges of N1-N3 are similar, and N1 is used most frequently, then N1-N3 can be merged into N1; in addition, the value ranges of N5 and N6 are similar, and N6 is used most frequently, then N5 and N6 can be merged into N6, and thus objection names N1, N4, N6 and N7 can be obtained. It should be noted that, in the case where there are at least two objection names, the meanings of each objection name in at least two objection names are different. Furthermore, based on each pool feature, at least one synonymous feature corresponding to each synonymous name is obtained, that is, the pool feature corresponding to at least one objection name is used as the objection feature, so that at least one objection feature can be obtained.

[0086] The above technical solution, by standardizing the synonymous features (i.e., attribute features) used in the same product identification process for each item in the item pool, effectively solves the problem of low accuracy in same product identification caused by the existence of synonyms in the feature names subjectively filled in by store merchants, thereby further improving the accuracy of same product identification.

[0087] Figure 4It is a flow chart of another same-product identification method provided in an embodiment of the present invention. This embodiment is optimized based on the above-mentioned technical solutions. In this embodiment, optionally, the above-mentioned same-product identification method also includes: taking the pool items in the item pool where the first item and the second item are located as nodes; for every two pool items in the item pool, according to the similarity of the values ​​corresponding to at least one attribute feature of the two pool items, obtaining the item similarity between the two pool items, and generating edges for connecting the nodes corresponding to the two pool items based on the item similarity; based on the nodes corresponding to each pool item in the item pool, and the edges corresponding to every two pool items, construct an item relationship network; based on the item relationship network, perform same-product identification for the first item and the second item. Among them, the explanations of the terms that are the same or corresponding to the above-mentioned embodiments are not repeated here.

[0088] See also Figure 4 The method of this embodiment may specifically include the following steps:

[0089] S310. For every two pool items in the item pool and each attribute feature of at least one attribute feature corresponding to the item pool, obtain the similarity of the values ​​corresponding to the attribute feature of the two pool items according to the pool attribute values ​​of the two pool items under the attribute feature.

[0090] Among them, for all the pool items in the item pool, any combination of these pool items in pairs can obtain one or more groups of two pool items. For example, continuing with the above example, when the item pool includes S1, S2, S3 and S4, any combination of these four pool items in pairs can obtain the following six groups of two pool items: S1 and S2, S1 and S3, S1 and S4, S2 and S3, S2 and S4, and S3 and S4. On this basis, for each two pool items in one or more groups of two pool items (for example, S1 and S2, S1 and S3, S1 and S4, S2 and S3, S2 and S4, and S3 and S4), the following operations are performed respectively:

[0091] For each attribute feature of at least one attribute feature, the similarity of the values ​​corresponding to the attribute feature of the two pool items is obtained based on the pool attribute value of one pool item under the attribute feature among the two pool items currently being processed and the pool attribute value of the other pool item under the attribute feature.

[0092] S320: Obtain the item similarity between the two pool items according to the similarity of the values ​​corresponding to at least one attribute feature of the two pool items.

[0093] In which, when the similarity of the values ​​corresponding to at least one attribute feature of the two pool items is obtained, the item similarity between the two pool items is obtained based on these similarities. In practical applications, the item similarity can be obtained in a variety of ways. Here, the importance of at least one attribute feature is taken as an example. For example, assuming that the similarities of the values ​​of the two items under n attribute features are f1, f2, ... f n , that is, the similarity of the pool attribute values ​​(such as 'first-level energy efficiency' and 'first-level energy consumption') corresponding to the first attribute feature (such as 'energy efficiency level') between the two items is f1, and the importance (i.e., weight) corresponding to each attribute feature is α1, α2, ... α n , then the overall similarity between the two items (i.e., item similarity) can be expressed as In another exemplary embodiment, the highest and lowest value similarities are removed from the value similarities corresponding to at least one attribute feature, and then the average value of the value similarities retained after the removal is used as the item similarity; and so on, which are not specifically limited here. The above example can be understood as obtaining the item similarity through a multi-channel recall scheme.

[0094] S330: The two pool items are respectively used as nodes, and based on the item similarity, edges are generated to connect the nodes corresponding to the two pool items.

[0095] Among them, the two pool items are respectively regarded as nodes. On this basis, based on the item similarity, an edge for connecting the nodes corresponding to the two pool items can be generated. For example, the size of the edge is represented by the item similarity. On this basis, the weight of the edge can also be represented by the importance corresponding to the item similarity.

[0096] S340: construct an item relationship network based on the nodes corresponding to the items in the item pool and the edges corresponding to every two items in the item pool.

[0097] Among them, when every two pool items in the item pool are processed through the above steps, at least two nodes and edges for connecting these nodes can be obtained. It should be noted that in actual applications, there are repeated pool items in every two pool items, for example, S1 exists in S1 and S2, S1 and S3, and S1 and S4, and the repeated pool items correspond to the same node.

[0098] At this point, an item relationship network can be constructed based on the nodes corresponding to each pool item in the item pool, and the edges corresponding to every two pool items. The item relationship network can be understood as the result of a networked representation of the relationship between pool items (such as item similarity).

[0099] S350: For a first item and a second item to be identified as the same item in the item pool, identify the first item and the second item as the same item based on the item relationship network.

[0100] Among them, since the item relationship network can be used to characterize the item similarity between any two pool items in the item pool, and the first item and the second item are both located in the item pool, the first item and the second item can be identified as the same item based on the item relationship network.

[0101] The technical solution of the embodiment of the present invention constructs an item relationship network for characterizing the relationship between any two pool items in the item pool, and performs same-identity identification for the first item and the second item in the item pool based on the item relationship network, thereby ensuring the accuracy of same-item identification.

[0102] On this basis, an optional technical solution is that the edge is an object edge, and the same-item identification method further includes:

[0103] Based on the similarity of the values ​​of the two pool items respectively corresponding to at least one attribute feature, an attribute edge is generated for connecting the nodes respectively corresponding to the two pool items;

[0104] Based on the nodes corresponding to each pool item in the item pool, and the edges corresponding to every two pool items, an item relationship network is constructed, including:

[0105] Based on the nodes corresponding to each pool item in the item pool, and the item edges and attribute edges corresponding to every two pool items, an item relationship network is constructed.

[0106] Among them, the edge generated based on the similarity of the items is used as the item edge. In addition, based on the similarity of the values ​​corresponding to at least one attribute feature of the two pool items, an attribute edge for connecting the nodes corresponding to the two pool items can be generated. The number of the attribute edges can be one, two or more. In practical applications, the attribute edges and attribute features can optionally correspond one to one, that is, each attribute feature corresponds to its own attribute edge. On this basis, an item relationship network is constructed based on nodes, item edges and attribute edges. The item relationship network can be used not only to characterize the item similarity between any two pool items, but also to characterize the similarity of their values ​​under the same feature attribute, characterizing the diversification of information, which helps to improve the accuracy of identifying the same product based on the item relationship network.

[0107] Figure 5It is a flow chart of another method for identifying similar products provided in an embodiment of the present invention. This embodiment is optimized based on the above-mentioned technical solutions. In this embodiment, optionally, based on the item relationship network, the first item and the second item are identified as similar products, including: performing community division on each node in the item relationship network to obtain at least one node cluster; based on the pool items represented by each node in the item relationship network, obtaining similar product clusters corresponding to at least one node cluster respectively, to obtain at least one similar product cluster; based on at least one similar product cluster, the first item and the second item are identified as similar products. Among them, the explanations of the terms that are the same or corresponding to the above-mentioned embodiments are not repeated here.

[0108] See also Figure 5 The method of this embodiment may specifically include the following steps:

[0109] S410. For every two pool items in the item pool and each attribute feature of at least one attribute feature corresponding to the item pool, obtain the value similarity of the two pool items corresponding to the attribute feature according to the pool attribute values ​​of the two pool items under the attribute feature.

[0110] S420: Obtain the item similarity between the two pool items according to the similarity of the values ​​corresponding to at least one attribute feature of the two pool items.

[0111] S430: The two pool items are respectively used as nodes, and edges for connecting the nodes corresponding to the two pool items are generated based on the item similarity.

[0112] S440: construct an item relationship network based on the nodes corresponding to the pool items in the item pool and the edges corresponding to every two pool items.

[0113] S450: Perform community division on each node in the item relationship network to obtain at least one node cluster.

[0114] Among them, each node in the item relationship network is divided into communities, so that the nodes with higher item similarity between the represented pool items can be divided into the same node cluster, and the nodes with lower item similarity between the represented pool items can be divided into different node clusters, thereby obtaining at least one node cluster. In practical applications, optionally, due to the lack of labels on whether two pool items are the same, the community division of each node can be achieved based on an unsupervised method, such as a graph segmentation algorithm or a graph clustering algorithm, wherein the graph segmentation algorithm refers to dividing the item relationship network (i.e., graph) into multiple clusters (i.e., subgraphs) of a preset size, so that the number of edges in the cluster is denser than the number of edges between clusters, such as spectral bisection or KL algorithm, etc.; the graph clustering algorithm refers to representing the graph as an adjacency matrix or a small matrix after dimensionality reduction such as matrix decomposition transformation, and then clustering, such as hierarchical clustering or spectral clustering, etc. Of course, the community division of each node can also be achieved by other methods, which are not specifically limited here.

[0115] S460. Based on the pool items represented by each node in the item relationship network, obtain clusters of the same product corresponding to at least one node cluster to obtain at least one cluster of the same product.

[0116] Among them, for each node cluster in at least one obtained node cluster, since the pool items represented by each node in the node cluster are known, a cluster of the same product corresponding to the node cluster can be obtained, and the item pool is divided into at least one cluster of the same product.

[0117] S470: For a first item and a second item to be identified as same items in the item pool, identify the first item and the second item as same items based on at least one same item cluster.

[0118] Among them, according to the characteristics of at least one of the above-mentioned node clusters, it can be inferred that the item similarity between pool items in the same same-product cluster is relatively high, and the item similarity between pool items in different same-product clusters is relatively low. Therefore, the first item and the second item can be identified as the same product based on these same-product clusters. For example, if the two items are in the same same-product cluster, they are determined to be the same product, otherwise they are not the same product.

[0119] It should be noted that in the process of dividing the same product clusters, a hierarchical network or clustering tree solution can be used to display it, and the importance of the attribute features or the threshold of the clustering distance can be manually modified, so that the output process of each same product cluster can be traced back to the dimension of the attribute features. In other words, the same product identification process can be visualized, and the judgment of whether the items in two pools are the same can be traced back to the dimension of the attribute features, making the same product identification process highly explainable.

[0120] The technical solution of the embodiment of the present invention is to divide the nodes in the item relationship network into communities, and then combine the pool items represented by these nodes to obtain at least one same-product cluster. Subsequently, the same-product identification can be performed based on these same-product clusters, thereby improving the convenience of same-product identification.

[0121] An optional technical solution, the above-mentioned same product identification method further includes:

[0122] For each of the at least one same-product cluster, determining whether the same-product cluster satisfies a preset same-product condition, wherein the preset same-product condition includes that each pool item in the same-product cluster is of the same product;

[0123] When there is a same-product cluster that does not meet the preset same-product condition in at least one same-product cluster, the communities of the nodes in the item relationship network are re-divided, or the same-product cluster that meets the preset same-product condition is selected from at least one same-product cluster for application.

[0124] Among them, in actual applications, there may be situations where the community division effect is not ideal, which means that the items in each pool within the same product cluster obtained at this time are not necessarily the same product. Therefore, in order to ensure the accuracy of subsequent same product identification, when at least one same product cluster is divided, the division result at this time can be evaluated, that is, to determine whether each same product cluster meets the preset same product condition, and the preset same product condition can be used to characterize that each pool item within the same product cluster is the same product. In the case where there are same product clusters that do not meet the preset same product conditions in at least one same product cluster, the communities of each node in the item relationship network can be re-divided, or, the same product clusters that meet the preset same product conditions can be selected from at least one same product cluster for application, so that the same product can be identified based on the processed same product cluster, thereby further improving the accuracy of same product identification, and reducing the impact on online business, with high comprehensiveness.

[0125] On this basis, optionally, it is possible to determine whether the same product cluster meets the preset same product conditions based on a variety of methods, such as a determination method based on labels, specifically, on some pool items that are manually labeled or have correct item models, based on indicators such as accuracy and / or recall rate for evaluation; another example is a determination method based on no labels, such as silhouette coefficient, Calinski-Harabaz index, Davidson-Bouldin Index or Dunn Validity Index. In addition, it can also be evaluated based on dimensions such as value similarity or acquisition cost. The specific implementation method is as follows:

[0126] Determine the acquisition cost of each pool item in the same product cluster, the cost indicator data under the preset aggregation indicator, and / or determine the value similarity of each pool item in the same product cluster under each attribute feature, the attribute indicator data under the preset clustering indicator; determine whether the same product cluster meets the preset same product conditions based on the cost indicator data and / or the attribute indicator data.

[0127] The preset aggregation index can be understood as a preset index used to characterize the degree of aggregation of the pool items in the same cluster in terms of acquisition cost, such as variance or error square sum. The preset clustering index can be understood as a preset index used to characterize the clustering of the pool items in the same cluster in terms of value similarity, such as sample silhouette coefficient or average diameter within the class. The above-mentioned preset clustering index can also be called clustering evaluation index or clustering assessment index.

[0128] Considering that the acquisition costs of each pool item of the same product are relatively close, and the value similarity is also relatively close, it is possible to determine whether the same product cluster meets the preset same product conditions based on the acquisition costs of each pool item in the same same product cluster, the cost indicator data under the preset aggregation indicator, and / or the value similarity of these pool items under each attribute feature, the attribute indicator data under the preset clustering indicator. The above technical solution evaluates the community division effect through the two dimensions of value similarity and acquisition cost, which helps to further improve the accuracy of subsequent same product identification.

[0129] In order to better understand the above-mentioned technical solutions, the following is an exemplary description of them with reference to specific examples. For example, in order to solve the problem of low accuracy in identifying the same product due to extremely irregular attribute values ​​of products across platforms and / or stores, see Figure 6 The same product identification scheme shown in the figure is based on the similarity of the values ​​of the same feature attributes to divide the community, specifically:

[0130] The attribute characteristics of each commodity in the same commodity pool are analyzed and processed, and the attribute characteristics with different meanings are determined from these attribute characteristics, so as to obtain at least one attribute characteristic corresponding to the commodity pool for application. The importance of each attribute characteristic currently applied is evaluated, so as to assign different weights to these attribute characteristics respectively. The value similarity between the attribute values ​​of each two commodities under the same attribute characteristic is calculated, so that multi-way recall can be performed based on the weight and the value similarity under different attribute characteristics to obtain the commodity similarity between each two commodities. Further, a multi-level link relationship (i.e., an item relationship network) between each two commodities is constructed based on the commodity similarity, value similarity and weight, and then the commodities are divided into multiple clusters of the same product based on the community discovery algorithm; finally, the division effect of the same product cluster is evaluated based on the value similarity or price dimension, and the same product clusters that do not meet the preset same product conditions are re-divided or the available same product clusters are selected from the full amount of same product clusters, so that the commodities can be managed in a standardized manner, and a more refined marketing strategy can be formulated to achieve business goals such as avoiding risks in advance and meeting the personalized needs of users.

[0131] Figure 7 This is a structural block diagram of a same-product identification device provided in an embodiment of the present invention, and the device is used to execute the same-product identification method provided in any of the above embodiments. The device and the same-product identification method of the above embodiments belong to the same inventive concept. For details not described in detail in the embodiment of the same-product identification device, refer to the embodiment of the same-product identification method. Figure 7 The device may specifically include: an attribute feature determination module 510, a value similarity determination module 520 and a same product identification module 530.

[0132] An attribute feature determination module 510 is used to determine at least one attribute feature matching the first object and the second object for the first object and the second object to be identified as the same object;

[0133] A value similarity determination module 520 is used to determine, for each attribute feature of the at least one attribute feature, a value similarity corresponding to the attribute feature according to a first attribute value of the first item under the attribute feature and a second attribute value of the second item under the attribute feature;

[0134] The same product identification module 530 is used to identify the first item and the second item as the same item according to the similarity of the values ​​corresponding to at least one attribute feature.

[0135] Optionally, the attribute feature determination module 510 may include:

[0136] a synonymous feature acquisition unit, configured to determine an item pool in which the first item and the second item are located, and acquire at least one synonymous feature corresponding to the item pool, wherein the at least one synonymous feature is determined according to a pool feature of each pool item in the item pool;

[0137] The attribute feature obtaining unit is used to use at least one synonymous feature as an attribute feature matching the first item and the second item, respectively, to obtain at least one attribute feature.

[0138] On this basis, optionally, the same product identification device further includes:

[0139] A pool attribute value acquisition module is used to acquire, for each pool item in the item pool, a feature name of a pool feature of the pool item and a pool attribute value of the pool item under the pool feature;

[0140] A value range obtaining module is used to obtain the value range corresponding to the feature name according to the values ​​of each pool attribute under the name feature for the name feature with the same feature name in each pool feature corresponding to the item pool;

[0141] A synonym name obtaining module is used for merging the feature names with similar value ranges in the at least two feature names corresponding to the item pool to obtain at least one synonym name;

[0142] The synonymous feature obtaining module is used to obtain synonymous features corresponding to at least one synonymous name based on each pool feature.

[0143] Optionally, the value similarity determination module 520 may include:

[0144] A second attribute value acquisition unit, used to acquire a first attribute value of the first item under the attribute feature, and a second attribute value of the second item under the attribute feature;

[0145] A second vector obtaining unit, used for inputting the first attribute value into the pre-trained model to obtain a first vector, and inputting the second attribute value into the pre-trained model to obtain a second vector;

[0146] The value similarity obtaining unit is used to determine the vector similarity between the first vector and the second vector, and use the vector similarity as the value similarity corresponding to the attribute feature.

[0147] Optionally, the same product identification module 530 is specifically used to:

[0148] Obtaining the importance level corresponding to at least one attribute feature;

[0149] The first object and the second object are identified as the same object according to the value similarity and importance respectively corresponding to the at least one attribute feature.

[0150] On this basis, optionally, the same product identification device further includes:

[0151] A training sample obtaining module is used to obtain, for each pool item in the item pool where the first item and the second item are located, the acquisition cost of the pool item and the pool attribute value of the pool item under each attribute feature as a set of training samples;

[0152] A cost prediction model obtaining module is used to train the original prediction model based on the training samples corresponding to each pool item to obtain a cost prediction model;

[0153] A feature indicator data obtaining module, used to obtain, for each attribute feature of at least one attribute feature, feature indicator data of the attribute feature obtained during the training process of the cost prediction model under a preset feature indicator;

[0154] The importance obtaining module is used to obtain the importance corresponding to at least one attribute feature according to the feature indicator data corresponding to at least one attribute feature.

[0155] Optionally, the same product identification device further includes:

[0156] A node obtaining module is used for taking the pool items in the item pool where the first item and the second item are located as nodes;

[0157] An edge generation module is used to obtain, for each two pool items in the item pool, the item similarity between the two pool items according to the similarity of the values ​​corresponding to at least one attribute feature of the two pool items, and to generate an edge for connecting the nodes corresponding to the two pool items based on the item similarity;

[0158] An item relationship network construction module is used to construct an item relationship network based on the nodes corresponding to each pool item in the item pool and the edges corresponding to every two pool items;

[0159] The same product identification module 530 may include:

[0160] The same product identification unit is used to identify the first item and the second item as the same item based on the item relationship network.

[0161] On this basis, optionally, the edge is an edge of an object, and the same-item identification device further includes:

[0162] An attribute edge generation module, used for generating an attribute edge for connecting nodes corresponding to two pool items respectively based on the similarity of values ​​corresponding to at least one attribute feature of the two pool items respectively;

[0163] Item relationship network building module, specifically used for:

[0164] Based on the nodes corresponding to each pool item in the item pool, and the item edges and attribute edges corresponding to every two pool items, an item relationship network is constructed.

[0165] Another optional same product identification unit is specifically used for:

[0166] Perform community division on each node in the item relationship network to obtain at least one node cluster;

[0167] Based on the pool items represented by each node in the item relationship network, obtain clusters of the same product corresponding to at least one node cluster, so as to obtain at least one cluster of the same product;

[0168] Based on at least one similar product cluster, similar product identification is performed on the first object and the second object.

[0169] On this basis, optionally, the same product identification device may include:

[0170] A same-product cluster identification module is used to determine, for each same-product cluster in at least one same-product cluster, whether the same-product cluster meets a preset same-product condition, wherein the preset same-product condition is that each pool item in the same-product cluster is of the same product;

[0171] The same product cluster processing module is used to re-divide the communities of each node in the item relationship network when there is a same product cluster that does not meet the preset same product conditions in at least one same product cluster, or to select the same product cluster that meets the preset same product conditions from at least one same product cluster for application.

[0172] On this basis, the optional same product cluster identification module may include:

[0173] An attribute index data determination unit, used to determine the acquisition cost of each pool item in the same product cluster, the cost index data under a preset aggregation index, and / or to determine the value similarity of each pool item in the same product cluster under each attribute feature, the attribute index data under a preset clustering index;

[0174] The same-product cluster identification unit is used to determine whether the same-product cluster meets the preset same-product condition according to the cost indicator data and / or the attribute indicator data.

[0175] The same product identification device provided by the embodiment of the present invention determines at least one attribute feature that matches the first and second items to be identified as the same product through an attribute feature determination module; determines the value similarity corresponding to the attribute feature for each attribute feature in at least one attribute feature according to the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature through a value similarity determination module; and performs same product identification for the first and second items according to the value similarity corresponding to at least one attribute feature through the same product identification module. The above device compares the attribute values ​​of all attribute features that match the first and second items for similarity. In this way, even if some of the attribute values ​​are not standardized, especially the attribute values ​​of key features are not standardized, the same product identification can be performed with reference to those standardized attribute values, thereby improving the accuracy of same product identification.

[0176] The same product identification device provided in the embodiment of the present invention can execute the same product identification method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0177] It is worth noting that in the above-mentioned embodiment of the same product identification device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific name of each functional unit is only for the convenience of distinguishing each other, and is not used to limit the scope of protection of the present invention.

[0178] Figure 8 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0179] like Figure 8As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0180] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0181] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a method for identifying the same product.

[0182] In some embodiments, the same product identification method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the same product identification method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the same product identification method in any other appropriate manner (for example, by means of firmware).

[0183] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0184] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0185] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer 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 foregoing.

[0186] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0187] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0188] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0189] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0190] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for identifying same product, characterized in that: include: For a first object and a second object to be identified as the same object, determining at least one attribute feature that matches the first object and the second object; For each attribute feature of the at least one attribute feature, determine the value similarity corresponding to the attribute feature according to the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature; The first object and the second object are identified as the same object according to the similarity of the values ​​corresponding to the at least one attribute feature.

2. The method according to claim 1, characterized in that Determining at least one attribute feature that matches the first item and the second item; Determine an item pool in which the first item and the second item are located, and obtain at least one synonymous feature corresponding to the item pool, wherein the at least one synonymous feature is determined according to a pool feature of each pool item in the item pool; The at least one synonymous feature is used as an attribute feature matching the first item and the second item, respectively, to obtain at least one attribute feature.

3. The method according to claim 2, characterized in that Also includes: For each pool item in the item pool, obtaining a feature name of a pool feature of the pool item and a pool attribute value of the pool item under the pool feature; For the name features with the same feature name in each pool feature corresponding to the item pool, the value range corresponding to the feature name is obtained according to the values ​​of each pool attribute under the name feature; For at least two feature names corresponding to the item pool, merging feature names with similar value ranges in the at least two feature names to obtain at least one synonymous name; Based on the respective pool features, synonymous features respectively corresponding to the at least one synonymous name are obtained.

4. The method according to claim 1, characterized in that: The determining, based on the first attribute value of the first item under the attribute feature and the second attribute value of the second item under the attribute feature, the similarity of the values ​​corresponding to the attribute feature includes: Obtaining a first attribute value of the first item under the attribute feature, and a second attribute value of the second item under the attribute feature; Inputting the first attribute value into a pre-trained model to obtain a first vector, and inputting the second attribute value into the pre-trained model to obtain a second vector; A vector similarity between the first vector and the second vector is determined, and the vector similarity is used as a value similarity corresponding to the attribute feature.

5. The method according to claim 1, characterized in that: The performing same-item identification on the first item and the second item according to the value similarity corresponding to the at least one attribute feature includes: Obtaining the importance level corresponding to each of the at least one attribute feature; The first object and the second object are identified as the same object according to the value similarity and the importance respectively corresponding to the at least one attribute feature.

6. The method according to claim 5, characterized in that Also includes: For each pool item in the item pool where the first item and the second item are located, the acquisition cost of the pool item and the pool attribute value of the pool item under each attribute feature are used as a set of training samples; Based on the training samples corresponding to each pool item, the original prediction model is trained to obtain a cost prediction model; For each attribute feature of the at least one attribute feature, acquiring feature indicator data of the attribute feature under a preset feature indicator obtained during the training process of the cost prediction model; According to the characteristic index data respectively corresponding to the at least one attribute feature, the importance degree respectively corresponding to the at least one attribute feature is obtained.

7. The method according to claim 1, characterized in that Also includes: The pool items in the item pool where the first item and the second item are located are used as nodes; For every two pool items in the item pool, obtaining the item similarity between the two pool items according to the similarity of the values ​​corresponding to the at least one attribute feature of the two pool items, and generating an edge for connecting the nodes corresponding to the two pool items based on the item similarity; Based on the nodes corresponding to the pool items in the item pool, and the edges corresponding to every two pool items, an item relationship network is constructed; The performing same-item identification on the first item and the second item according to the value similarity corresponding to the at least one attribute feature includes: Based on the item relationship network, the first item and the second item are identified as the same item.

8. The method according to claim 7, characterized in that The edge is an item edge, and the method further includes: Based on the similarity of the values ​​of the two pool items respectively corresponding to the at least one attribute feature, generating an attribute edge for connecting the nodes respectively corresponding to the two pool items; The item relationship network is constructed based on the nodes corresponding to the pool items in the item pool and the edges corresponding to every two pool items, including: Based on the nodes corresponding to the pool items in the item pool, and the item edges and attribute edges corresponding to every two pool items, an item relationship network is constructed.

9. The method according to claim 7, characterized in that: The performing same-item identification on the first item and the second item based on the item relationship network includes: Performing community division on each node in the item relationship network to obtain at least one node cluster; Based on the pool items represented by each node in the item relationship network, obtain clusters of the same product corresponding to the at least one node cluster, so as to obtain at least one cluster of the same product; Based on the at least one same-product cluster, the first object and the second object are identified as being of the same product.

10. The method according to claim 9, characterized in that Also includes: For each of the at least one same product cluster, determining whether the same product cluster satisfies a preset same product condition, wherein the preset same product condition includes that each pool item in the same product cluster is of the same product; In the case that there is a same-product cluster that does not meet the preset same-product condition in the at least one same-product cluster, the communities of the nodes in the item relationship network are re-divided, or the same-product cluster that meets the preset same-product condition is selected from the at least one same-product cluster for application.

11. The method according to claim 10, characterized in that The determining whether the same product cluster meets a preset same product condition includes: Determine the acquisition cost of each pool item in the same product cluster, the cost indicator data under the preset aggregation indicator, and / or determine the value similarity of each pool item in the same product cluster under each attribute feature, the attribute indicator data under the preset clustering indicator; It is determined whether the same-product cluster meets a preset same-product condition according to the cost indicator data and / or the attribute indicator data.

12. A device for identifying same product, characterized in that: include: An attribute feature determination module, for determining, for a first object and a second object to be identified as same-items, at least one attribute feature matching the first object and the second object; a value similarity determination module, configured to determine, for each attribute feature of the at least one attribute feature, a value similarity corresponding to the attribute feature according to a first attribute value of the first item under the attribute feature and a second attribute value of the second item under the attribute feature; The same product identification module is used to identify the first item and the second item as the same item according to the similarity of the values ​​corresponding to the at least one attribute feature.

13. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor executes the same product identification method as described in any one of claims 1-11.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the same product identification method as described in any one of claims 1-11 when executed.

Images