Data search methods, devices, equipment and storage media

Abstract

The embodiment of the present invention provides a data search method, device, equipment and storage medium, the method comprising: receiving data search information corresponding to a data category input by a user; if there is no reference query information matching the data search information in the question-and-answer library corresponding to the data category, determining attribute description information matching the data search information from the data attribute library corresponding to the data category; according to the attribute value screening condition corresponding to the attribute description information, screening out the data set corresponding to the attribute value screening condition from the candidate data set corresponding to the data category and outputting it. Through this scheme, when the data search information input by the user is relatively short and cannot directly locate the data that satisfies the attribute condition that the user wants to search, data that meets the user's needs can be output for the user.

Description

Technical Field

[0001] The present invention relates to the field of Internet technology, and in particular to a data search method, device, equipment and storage medium. Background Art

[0002] With the development of Internet technology, people can obtain all kinds of items and information they need through the Internet without leaving home. For example, they can buy various goods online through various shopping apps.

[0003] Assume that users can purchase a certain category of goods through an APP, such as phone recharge cards, etc. In actual applications, when purchasing goods, users may input search statements that reflect their purchase intentions. However, the search statements may be very short sentences without context information. For example, the search statements entered by users are: cheap, best, fast, cost-effective, etc. Such search statements make the APP confused and unable to accurately understand the user's true intentions, and thus unable to accurately recommend goods that meet the user's needs. Summary of the invention

[0004] The embodiments of the present invention provide a data search method, apparatus, device and storage medium, which can provide users with data that meets their needs.

[0005] In a first aspect, an embodiment of the present invention provides a data search method, the method comprising:

[0006] Receiving data search information corresponding to the data category input by the user;

[0007] If there is no reference query information matching the data search information in the question-answer library corresponding to the data category, determining attribute description information matching the data search information from the data attribute library corresponding to the data category;

[0008] According to the attribute value screening condition corresponding to the attribute description information, screening out a data set corresponding to the attribute value screening condition from the candidate data sets corresponding to the data category;

[0009] Output the dataset.

[0010] In a second aspect, an embodiment of the present invention provides a data search device, the device comprising:

[0011] A receiving module, used for receiving data search information corresponding to a data category input by a user;

[0012] a processing module, configured to determine attribute description information matching the data search information from a data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question-answer library corresponding to the data category; and to filter out a data set corresponding to the attribute value filtering condition from candidate data sets corresponding to the data category according to an attribute value filtering condition corresponding to the attribute description information;

[0013] An output module is used to output the data set.

[0014] In a third aspect, an embodiment of the present invention provides an electronic device, comprising: a memory and a processor; wherein the memory stores executable code, and when the executable code is executed by the processor, the processor can at least implement the data search method described in the first aspect.

[0015] In a fourth aspect, an embodiment of the present invention provides a non-temporary machine-readable storage medium having executable code stored thereon. When the executable code is executed by a processor of an electronic device, the processor can at least implement the data search method described in the first aspect.

[0016] In a fifth aspect, an embodiment of the present invention provides a data search method, the method comprising:

[0017] In response to an image input by a user, identifying data search information corresponding to a data category from the image;

[0018] If there is no reference query information matching the data search information in the question-answer library corresponding to the data category, determining attribute description information matching the data search information from the data attribute library corresponding to the data category;

[0019] According to the attribute value screening condition corresponding to the attribute description information, screening out a data set corresponding to the attribute value screening condition from the candidate data sets corresponding to the data category;

[0020] Output the data set.

[0021] In a sixth aspect, an embodiment of the present invention provides a data search device, the device comprising:

[0022] an identification module, for identifying data search information corresponding to a data category from an image input by a user;

[0023] an acquisition module, configured to determine attribute description information matching the data search information from a data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question-answer library corresponding to the data category; and to filter out a data set corresponding to the attribute value filtering condition from candidate data sets corresponding to the data category according to an attribute value filtering condition corresponding to the attribute description information;

[0024] An output module is used to output the data set.

[0025] In the seventh aspect, an embodiment of the present invention provides an electronic device, comprising: a memory and a processor; wherein the memory stores executable code, and when the executable code is executed by the processor, the processor can at least implement the data search method as described in the fifth aspect.

[0026] In an eighth aspect, an embodiment of the present invention provides a non-temporary machine-readable storage medium having executable code stored thereon. When the executable code is executed by a processor of an electronic device, the processor can at least implement the data search method as described in the fifth aspect.

[0027] In a ninth aspect, an embodiment of the present invention provides a data search method, the method comprising:

[0028] In response to a video input by a user, identifying data search information corresponding to a data category from the video;

[0029] If there is no reference query information matching the data search information in the question-answer library corresponding to the data category, determining attribute description information matching the data search information from the data attribute library corresponding to the data category;

[0030] According to the attribute value screening condition corresponding to the attribute description information, screening out a data set corresponding to the attribute value screening condition from the candidate data sets corresponding to the data category;

[0031] Output the dataset.

[0032] In a tenth aspect, an embodiment of the present invention provides a data search device, the device comprising:

[0033] an identification module, for identifying data search information corresponding to a data category from a video input by a user;

[0034] an acquisition module, configured to determine attribute description information matching the data search information from a data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question-answer library corresponding to the data category; and to filter out a data set corresponding to the attribute value filtering condition from candidate data sets corresponding to the data category according to an attribute value filtering condition corresponding to the attribute description information;

[0035] An output module is used to output the data set.

[0036] In the eleventh aspect, an embodiment of the present invention provides an electronic device, comprising: a memory and a processor; wherein the memory stores executable code, and when the executable code is executed by the processor, the processor can at least implement the data search method as described in the ninth aspect.

[0037] In the twelfth aspect, an embodiment of the present invention provides a non-temporary machine-readable storage medium having executable code stored thereon. When the executable code is executed by a processor of an electronic device, the processor can at least implement the data search method as described in the ninth aspect.

[0038] In an embodiment of the present invention, when a user outputs data search information for a certain data category, if the data search information is a default short message and there is no matching reference query information in the question-and-answer library corresponding to the data category, attribute description information matching the data search information is determined from the data attribute library corresponding to the data category to clarify what attribute characteristics the user wants to search for data with, and then, based on the attribute value filtering condition corresponding to the matched attribute description information, a data set corresponding to the attribute value filtering condition is filtered out from the candidate data set corresponding to the data category and output to the user, thereby ensuring that the data output to the user is consistent with the user's true intention. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] 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 some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 A flowchart of a data attribute library construction process provided by an embodiment of the present invention;

[0041] Figure 2 A schematic diagram of a data attribute library construction result provided by an embodiment of the present invention;

[0042] Figure 3 A flow chart of a data search method provided by an embodiment of the present invention;

[0043] Figure 4 A schematic diagram of a scenario for executing a data search method provided by an embodiment of the present invention;

[0044] Figure 5 A flowchart of another data search method provided by an embodiment of the present invention;

[0045] Figure 6 A flowchart of another data search method provided by an embodiment of the present invention;

[0046] Figure 7 A schematic diagram of the structure of a data search device provided by an embodiment of the present invention;

[0047] Figure 8 For Figure 7 A schematic diagram of the structure of an electronic device corresponding to the data search device provided in the illustrated embodiment;

[0048] Fig. 9 A schematic diagram of the structure of another data search device provided by an embodiment of the present invention;

[0049] Fig.10 For Fig. 9 A schematic diagram of the structure of an electronic device corresponding to the data search device provided in the illustrated embodiment;

[0050] Fig.11 A schematic diagram of the structure of another data search device provided by an embodiment of the present invention;

[0051] Fig.12 For Fig.11 A schematic structural diagram of an electronic device corresponding to the data search device provided in the illustrated embodiment. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution 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 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 are within the scope of protection of the present invention.

[0053] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings, and "multiple" generally includes at least two.

[0054] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0055] In addition, the step sequence in the following method embodiments is only an example and not a strict limitation.

[0056] The data search method provided in the embodiment of the present invention can be executed by an electronic device, which can be a terminal device such as a PC, a laptop, a smart phone, etc., or a server. The server can be a physical server including an independent host, or a virtual server, or a cloud server.

[0057] Before introducing the execution process of the data search method provided by the embodiment of the present invention, the overall idea of ​​the data search method is first introduced. The data search process can be regarded as a human-computer question-answering dialogue process.

[0058] Taking the e-commerce scenario as an example, a question-and-answer library corresponding to the product category will be established in advance. In simple terms, the question-and-answer library will contain reference query information reflecting the user's shopping intention and corresponding reply data, which is reflected in individual products. Therefore, when the user outputs search information for the product category, the question-and-answer library is searched for reference query information matching the search information. When there is matching reference query information, the reply corresponding to the reference query information is output to the user. Among them, the reference query information will clearly indicate the attributes and attribute values ​​of the product, so that the reference query information will clearly point to one or more specific products.

[0059] However, when a user outputs search information for a certain product category, in fact, only a very short search statement may be given, and there is no context. For example, when a user buys a phone recharge card, he will enter short texts such as cheap, unavailable, fast, best, and cost-effective. This type of description is often confusing, that is, the search information does not clearly indicate the attributes and attribute values ​​of the product to be searched, and often only gives keywords corresponding to the product attributes. The e-commerce platform may not be able to provide an accurate list of products. This is because, on the one hand, the above-mentioned question and answer library cannot match the reference query information that matches this search information, and on the other hand, some attributes of the product are dynamically changing. For example, if a user enters "cheap", how much is a certain product considered "cheap"? It may depend on a series of factors such as the merchant's location, prices, product materials, production date, policies, logistics, etc. Therefore, this short search statement cannot be clearly mapped to a specific product.

[0060] In order to cope with the situation where the user inputs the above short attributes and attribute values ​​that do not indicate the product to be searched, the embodiment of the present invention proposes the idea of ​​building a data attribute library. Among them, corresponding to the above e-commerce scenario, the data attribute library can be called a product attribute library. The product attribute library corresponds to the product category, and contains the correspondence between product attributes, attribute description information, and attribute value screening conditions.

[0061] In summary, the data search information in this article may have the following characteristics: the data search information may contain keywords corresponding to data attributes, but the data search information does not contain data attribute values, so that the data set that needs to be fed back to the user cannot be clearly located based on the literal meaning of the data search information. In other words, the data search information has a "fuzzy" characteristic and cannot directly correspond to the attribute value of a certain data attribute.

[0062] Of course, the above description is only based on the e-commerce scenario. In fact, the solution provided by the embodiment of the present invention can be applied to a variety of different application scenarios, including but not limited to e-commerce scenarios, such as article search scenarios, video search scenarios, etc. Therefore, the data mentioned below can be commodities, articles, videos, etc.

[0063] Next, we will first introduce how to build a data attribute library, and then introduce how to perform data search and processing based on the data attribute library.

[0064] Figure 1 A flowchart of a data attribute library construction process provided by an embodiment of the present invention is as follows: Figure 1 As shown, the method comprises the following steps:

[0065] 101. Obtain multiple search corpora and multiple data attributes corresponding to the data category.

[0066] 102. Determine attribute description information corresponding to each of the plurality of data attributes according to the plurality of search corpora.

[0067] 103. Obtain attribute value filtering conditions corresponding to each of multiple data attributes.

[0068] 104. Generate a data attribute library according to the multiple data attributes, the attribute description information corresponding to each of the multiple data attributes, and the attribute value screening condition.

[0069] In the e-commerce scenario, the above data categories are commodity categories, such as phone card categories, dress categories, mobile phone categories, etc. In the article search scenario, the above categories are, for example, financial categories, entertainment categories, etc.

[0070] It is understandable that the data attributes corresponding to different data categories may not be the same. For example, the products corresponding to the mobile phone category are mobile phones of various brands, and their data attributes, namely product attributes, may include brand, model, operating system type, screen size, number of cameras and resolution, etc. The products corresponding to the dress category are various dresses, and their data attributes, namely product attributes, may include brand, color, length, fabric, size, etc.

[0071] In the embodiment of the present invention, for any data category, the data attributes contained in each data under the data category can be counted to obtain the data attributes corresponding to the data category. The above multiple data attributes can be all data attributes corresponding to the data category.

[0072] For any data category, the multiple search corpora corresponding to the data category may be all or part of the search information triggered by many users on the data category in history. The search information may be in text form or in voice form. When in voice form, it needs to be converted into text to obtain the search corpus in text form.

[0073] In order to balance the amount of data calculation and accuracy, optionally, multiple search corpora corresponding to any data category can be obtained in the following way:

[0074] Obtain the search corpus corresponding to the data category;

[0075] Perform word segmentation on each search corpus contained in the search corpus set;

[0076] Determine a preset number of target words with the highest word frequency;

[0077] Search corpora containing any target word among the multiple target words are screened out from the search corpus set as multiple search corpora corresponding to the data category.

[0078] The search corpus may be composed of all search information triggered by many users on the data category in history.

[0079] After word segmentation processing is performed on each search corpus contained in the search corpus, a word list contained in the search corpus can be obtained. For each word in the word list, the word frequency of the word is determined according to the number of occurrences of the word in the search corpus. Therefore, based on the word frequency of each word, the N words with the highest word frequency can be determined, where N is greater than 1. Finally, the search corpus containing any one of the N words is filtered out, and the filtered search corpus is one of the multiple search corpora corresponding to the data category.

[0080] After obtaining multiple search corpora and multiple data attributes corresponding to a certain data category, attribute description information corresponding to each of the multiple data attributes is determined based on the multiple search corpora, wherein, for any data attribute, the attribute description information corresponding to the data attribute reflects what search keywords many users are accustomed to using under the dimension of the data attribute.

[0081] Optionally, determining the attribute description information corresponding to each of the multiple data attributes according to the multiple search corpora may be implemented as follows:

[0082] Determine the similarity between the encoding vectors corresponding to each of the plurality of search corpora and the encoding vectors corresponding to each of the plurality of data attributes. For any of the plurality of data attributes, determine that the attribute description information corresponding to the any of the data attributes includes a search corpus whose similarity with the encoding vector of the any of the data attributes is greater than a set threshold. Specifically, the search keyword contained in the search corpus may also be used as the attribute description information corresponding to the data attribute.

[0083] In practical applications, a coding model can be trained in advance based on the aforementioned search corpus, so as to use the coding model to obtain coding vectors corresponding to each of the multiple search corpora and coding vectors corresponding to each of the multiple data attributes. The search corpus will contain a large number of search corpora that clearly indicate data attributes, and these search corpora will also contain attribute description keywords corresponding to the data attributes. Therefore, based on the coding model, the distance between the coding vectors of the search corpora describing the same data attribute can be closer, the distance between the coding vectors of the search corpora describing different data attributes can be farther, and the distance between the coding vectors of the search corpora describing the same data attribute and the coding vector of the data attribute can be closer.

[0084] For example, suppose a search corpus is: I want to find cheap. Another search corpus is: cheap. The data attribute is price. The encoding based on the above encoding model will make the encoding vector of "price" closer to the encoding vector of "cheap".

[0085] For the above-mentioned multiple search corpora and multiple data attributes, after obtaining the corresponding encoding vectors based on the encoding model, for each data attribute, the similarity between its encoding vector and the encoding vector of each search corpus can be calculated, and the similarity is obtained by calculating the cosine distance, Euclidean distance, etc. between the two encoding vectors. When the similarity between the encoding vector of a search corpus and the encoding vector of the data attribute is greater than a set threshold, it is considered that the search corpus or the keywords in the search corpus are a type of attribute description information corresponding to the data attribute.

[0086] In addition, optionally, in order to ensure that the attribute description information corresponding to the data attribute can cover the various expression habits of users, after determining a certain search corpus as the attribute description information corresponding to a certain data attribute, the search corpus can also be synonym expanded (it can be a synonym expansion of the keywords contained in the search corpus), so that the attribute description information corresponding to the data attribute also includes the information obtained after the synonym expansion. For example, assuming that a certain data attribute is price, the corresponding attribute description information is: cheap. By performing synonym expansion on "cheap", synonyms such as cheap, cost-effective, and inexpensive can be obtained.

[0087] The means of synonym expansion may be, for example, to collect various document data publicly available on the Internet, perform word segmentation processing on them, and for a certain data attribute, calculate the similarity between each word and the data attribute or the existing attribute description information corresponding to the data attribute, and determine the words with a similarity greater than a set threshold as a piece of attribute description information corresponding to the data attribute.

[0088] In addition, it is worth noting that for some data attributes, the corresponding multiple attribute description information may have the following situation: the semantics of some attribute description information is completely opposite to the semantics of other attribute description information. For example, the attribute description information corresponding to the data attribute of price may include the following attribute description information: cheap, cost-effective, cheap, expensive, and high-priced. Among them, cheap, cost-effective, and cheap have one semantics, while expensive and high-priced have another semantics.

[0089] Based on this, optionally, the multiple pieces of attribute description information corresponding to any data attribute may also be processed as follows: clustering is performed on the multiple pieces of attribute description information to obtain multiple categories of attribute description information corresponding to the data attribute through clustering.

[0090] Among them, it can be understood that the multiple categories here often represent semantic categories, that is, attribute description information representing different semantics corresponding to each data attribute is obtained. In addition, after clustering multiple attribute description information corresponding to a data attribute, only one category may be obtained. In the process of clustering, the corresponding clustering parameters can be set in advance for each data attribute. For example, for the data attribute of price, the similarity threshold between the encoding vectors of the attribute description information is pre-set, so that the attribute description information whose similarity between the encoding vectors of the attribute description information meets the similarity threshold is clustered into one category. For example, for the data attribute of usage time, different time ranges can be pre-set, so that the attribute description information corresponding to a certain time range is clustered into one category.

[0091] After obtaining the attribute description information corresponding to each of the multiple data attributes, the multiple data attributes and all or part of the attribute description information corresponding to each data attribute can be output to each data provider (such as a merchant) or a manager (such as an e-commerce platform maintenance personnel, news website maintenance personnel), and the data provider or manager sets the attribute value filtering conditions corresponding to each data attribute.

[0092] It is understandable that if a data attribute corresponds to multiple categories of attribute description information, an attribute value screening condition corresponding to each category of attribute description information is set.

[0093] Taking the e-commerce scenario as an example, suppose that a data attribute under a certain product category is price, and the price corresponds to two types of attribute description information: "cheap" and "expensive". For any merchant corresponding to the product category, the merchant can set the attribute value filtering condition corresponding to the attribute description information of "cheap" according to its own situation: (0,100] or (0,top*60%]. Among them, (0,100] means that the merchant sets the product under the product category for sale, and if the price is less than or equal to 100 yuan, it is cheap. (0,top*60%] means that the merchant sets multiple products under the product category for sale, and if the price is less than or equal to 60% of the highest price (i.e., top) among these multiple products, it is cheap.

[0094] It can be seen from this that each merchant can set attribute value filtering conditions corresponding to each attribute description information based on its own situation. Based on the setting of attribute description information and attribute value filtering conditions corresponding to each data attribute, the relatively vague data search information input by the user can be mapped to attribute value filtering conditions that can be clearly quantified, thereby assisting the user's data search process.

[0095] For ease of understanding, the following Figure 2 To illustrate the composition of the data attribute library obtained through the above-mentioned construction process.

[0096] exist Figure 2 In the example, taking the commodity category of phone cards as an example, it is assumed that the data attributes include three attributes: price, speed, and usage time. Each data attribute corresponds to multiple attribute description information, and each data attribute corresponds to an attribute value filtering condition.

[0097] The following examples illustrate how to perform data search based on the data attribute library constructed in the above examples.

[0098] Figure 3 A flowchart of a data search method provided by an embodiment of the present invention is shown in FIG. Figure 3 As shown, the method comprises the following steps:

[0099] 301. Receive data search information corresponding to a data category input by a user.

[0100] In actual applications, the user can input the data search information in text form or voice form. When the data search information is spoken in voice form, the voice can be converted into text to obtain the data search information in text form.

[0101] 302. If there is no reference query information matching the data search information in the question and answer library corresponding to the data category, determine the attribute description information matching the data search information from the data attribute library corresponding to the data category, and based on the attribute value filtering condition corresponding to the attribute description information, filter out the data set corresponding to the attribute value filtering condition from the candidate data sets corresponding to the data category.

[0102] 303. Output the data set.

[0103] As mentioned above, the question and answer library corresponding to a certain data category contains reference query information and response data corresponding to the reference query information. If reference query information matching the data search information currently input by the user can be found in the question and answer library, the response data corresponding to the reference query information can be directly output to the user.

[0104] In an optional embodiment, the question-and-answer database can be implemented as a knowledge base corresponding to the above data category, or a knowledge graph. The knowledge base includes entities and entity relationships corresponding to the data category. Therefore, if the data search information input by the user does not contain an entity in the knowledge base, it is determined that there is no reference query information matching the data search information in the knowledge base.

[0105] Taking the e-commerce scenario as an example, the entities in the knowledge base corresponding to a certain product category can include products and product attribute values. In this case, the entity relationship between the product entity and the product attribute value entity can be an attribute. Of course, the entity can also be a product. In this case, each product can be associated with multiple attribute values, and the entity relationship between two product entities can be the relationship between the two products.

[0106] The concept of knowledge graph is not elaborated in this article as it belongs to the existing technology.

[0107] In this embodiment, it is only emphasized that when the above-mentioned question-and-answer database is implemented as a knowledge base, it is possible to determine whether a data set corresponding to the data search information currently input by the user can be found in the knowledge base by identifying whether the data search information currently input by the user contains entities in the knowledge base. In practical applications, a named entity recognition model can be used to identify whether the data search information contains entities in the knowledge base.

[0108] When there is no reference query information matching the data search information in the question and answer database, the aforementioned data attribute database corresponding to the data category is used to perform data search processing.

[0109] First, determine the attribute description information that matches the data search information input by the user from the data attribute library. Specifically, you can first query whether the data search information exists in the data attribute library, that is, whether there is an attribute description information in the data attribute library that is consistent with the data search information. If the data search information does not exist in the data attribute library, encode the data search information to obtain the encoding vector corresponding to the data search information, and then determine the similarity between the encoding vector corresponding to each attribute description information in the data attribute library and the encoding vector corresponding to the data search information. Finally, determine that the attribute description information whose similarity with the encoding vector corresponding to the data search information is greater than the set threshold is the attribute description information that matches the data search information.

[0110] After obtaining the attribute description information matching the data search information, according to the attribute value screening condition corresponding to the attribute description information, a data set corresponding to the attribute value screening condition is screened out from the candidate data sets corresponding to the data category.

[0111] For ease of understanding, the following Figure 4 The illustrated embodiment is used to exemplarily illustrate the execution process of the data search method.

[0112] exist Figure 4 In the example of e-commerce, there are many commodity categories maintained in the e-commerce server. Suppose the user currently triggers the phone card category. Suppose the data search information entered by the user is: I want something fast. Figure 4As can be seen from the data attribute library shown in the figure, after receiving the data search information of the user, the user terminal determines that the similarity between the encoding vector of the data search information and the encoding vector of each attribute description information meets the requirements by calculating the similarity between the data search information and the attribute description information of "fast network speed" in the data attribute library. Therefore, the attribute value screening condition corresponding to the attribute description information: product generation (generation)>3G is used to screen out the phone cards that meet the attribute value screening condition from the phone cards stored in the server, such as Figure 4 In the figure, there are card 1, card 2, card 3, ... card n.

[0113] It is worth noting that, while each merchant can set the attribute value filtering conditions corresponding to each data attribute (or each attribute description information) according to its own situation, the attribute value filtering conditions set by each merchant for the same data attribute may not be the same. Therefore, the number of attribute value filtering conditions corresponding to the matched attribute description information may be multiple. At this time, each merchant can filter their respective products according to the attribute value filtering conditions set by each merchant to filter out the products that meet the attribute value filtering conditions set by themselves.

[0114] In an optional embodiment, after the data set corresponding to the user's data search information is screened out, the data contained in the data set can be displayed in sequence in a random order. In another optional embodiment, the display order of the data in the data set can also be determined according to the attribute value corresponding to each data in the data set. For example, in the order of attribute values ​​from low to high or from high to low.

[0115] In addition, taking the e-commerce scenario as an example, as mentioned above, the attribute value screening conditions set by each merchant for the same type of attribute description information may not be the same. Taking the price attribute as an example, for the attribute description information such as "cheap, cheap, cost-effective", it is assumed that the attribute value screening conditions set by most merchants are (0,100], and the attribute value screening conditions set by some merchants are [200,300]. At this time, it can be considered that the condition [200,300] is abnormal, and the abnormal condition can be eliminated during the data screening process. Alternatively, in the process of sorting the data in the data set, the abnormal attribute values ​​are processed: abnormal attribute values ​​in the data set are detected; data corresponding to the abnormal attribute values ​​are eliminated, or data corresponding to the abnormal attribute values ​​are arranged after data corresponding to non-abnormal attribute values. Optionally, when performing abnormal attribute value detection, the attribute values ​​contained in the data set can be fitted into a Gaussian distribution curve, and attribute values ​​far away from the Gaussian distribution curve are determined as abnormal attribute values.

[0116] The above embodiments describe data search based on a data attribute library. In practical applications, other factors may also be obtained to assist in data search.

[0117] For example, in an optional embodiment, the user's characteristic information can be obtained, so that after a data set (for ease of description, referred to as the first data set) is obtained through the above-mentioned attribute value filtering conditions, a data set (referred to as the second data set) that matches the user's characteristic information can be further filtered out from the data set.

[0118] Optionally, the characteristic information of the user includes at least one of the following: location information, historical search behavior. The location information is used to remove data in the first data set that does not conform to the location information. Taking the e-commerce scenario as an example, the historical search behavior reflects which specific product the user has searched for or purchased, so that the products in the first data set can be filtered according to the product information, such as filtering out products of the same brand.

[0119] For example, in the scenario of purchasing phone cards mentioned above, a user in Beijing enters the data search information "cheap". After obtaining the above-mentioned first data set according to the data attribute library, the data whose merchant addresses are in Beijing or the applicable addresses are in Beijing can be filtered out to form the second data set.

[0120] In summary, when a user inputs data search information for a certain data category, if the data search information is a default short message and cannot accurately locate the specific attribute conditions that the data that the user wants to search for meets, the solution provided by the embodiment of the present invention can be used to recommend data sets that meet the user's needs. The key point of the solution is to map the short information input by the user to the attribute value screening conditions for screening data, so as to quantify the user's search intention.

[0121] Figure 5 A flowchart of another data search method provided by an embodiment of the present invention, such as Figure 5 As shown, the data search method may include the following steps:

[0122] 501. In response to an image input by a user, identify data search information corresponding to a data category from the image.

[0123] 502. If reference query information matching the data search information does not exist in the question-and-answer library corresponding to the data category, attribute description information matching the data search information is determined from the data attribute library corresponding to the data category.

[0124] 503. According to the attribute value screening condition corresponding to the attribute description information, a data set corresponding to the attribute value screening condition is screened out from the candidate data sets corresponding to the data category.

[0125] 504. Output the data set.

[0126] In this embodiment, the method for a user to input data search information is not limited to the text and voice methods described in the other embodiments above, and data search information can also be provided by inputting images.

[0127] Taking the e-commerce scenario as an example, the above data category is a certain commodity category. The user clicks on the commodity category to browse the many commodities contained in the commodity category. When browsing a certain commodity, the user sees the advantages and characteristics of the commodity. If one of the advantages and characteristics is of interest to the user, the user can capture the text containing the description of the advantages and characteristics by taking a screenshot. Then, the user can trigger the following search operation: upload the captured image in the search interface corresponding to the above commodity category. Based on this, the execution of the above multiple steps provided in this embodiment is triggered. Among them, the above advantages and characteristics are often keywords corresponding to certain data attributes (commodity attributes in the e-commerce scenario). For example, the commodity category is phone cards, and the advantages and characteristics are: cheap and fast network speed.

[0128] Specifically, after receiving the image input by the user, character recognition processing can be performed on the image to obtain the text contained in the image, such as recognizing the text contained in the image through optical character recognition (OCR) technology. Then, keywords corresponding to data attributes are extracted from all the recognized texts as data search information.

[0129] In actual applications, the images provided by users may contain some interference texts, which should not be included in the data search information. Therefore, it is necessary to extract the texts that can be used as data search information from all the recognized texts. The extracted texts are keywords corresponding to the data attributes, that is, the texts describing the advantages and characteristics as described above. Optionally, when the user captures the keywords describing the above advantages and characteristics, the keywords often occupy a major position in the captured image, and generally speaking, the font size of the keywords is larger than other interference texts. Therefore, the keywords used as data search information can be extracted based on these image features.

[0130] In the above example, the data category (such as product category) is actively selected by the user from the category list, but, optionally, the data category that the user currently wants to search can also be obtained through other methods. The specific implementation method is as follows: the image input by the user includes an object, so the category corresponding to the object can be determined as the data category that the user wants to search for data.

[0131] In actual applications, the user can provide an image that contains both the above keywords and the object. The user can also provide two images, one containing the above keywords and the other containing the object.

[0132] For images containing objects, the categories of the objects contained in the images can be identified through various existing image classification schemes, and the categories are used as data categories.

[0133] Of course, it is understandable that the method of determining data categories is not limited to this. For example, an image provided by a user contains both the above-mentioned keywords and the name of an object (such as a phone card). Based on the extraction of the object name, the data category can also be determined.

[0134] In this embodiment, the execution process of steps 502-503 can refer to the relevant descriptions in the aforementioned other embodiments, and will not be repeated here.

[0135] Figure 6 A flowchart of another data search method provided by an embodiment of the present invention is as follows: Figure 6 As shown, the method may include the following steps:

[0136] 601. In response to a video input by a user, identify data search information corresponding to a data category from the video.

[0137] 602. If reference query information matching the data search information does not exist in the question-and-answer library corresponding to the data category, attribute description information matching the data search information is determined from the data attribute library corresponding to the data category.

[0138] 603. According to the attribute value screening condition corresponding to the attribute description information, a data set corresponding to the attribute value screening condition is screened out from the candidate data sets corresponding to the data category.

[0139] 604. Output the data set.

[0140] In this embodiment, the method for users to input data search information is not limited to the text, voice, and image methods described in the other embodiments above, and data search information can also be provided by inputting videos.

[0141] Take the e-commerce scenario as an example, the above data category is a certain commodity category. Suppose the user already owns a certain commodity, but the commodity is broken or the user wants to replace it with a new one, then the user will have the need to buy another commodity of this type (that is, want to buy commodities under the same commodity category, which may not be exactly the same as the original commodity). At this time, the user can shoot a short video of the broken commodity, and can also express the intention to buy by voice during the shooting process. Among them, the purchase intention can be represented by keywords that reflect the attributes of the commodity. For example, if the user wants to buy a mobile phone, he will say something like "I want to buy a large-screen and light-weight one".

[0142] After taking the video, the user may trigger the following search operation: upload the video in the search interface corresponding to the product category or in the homepage interface of the shopping app, thereby triggering the execution of the above steps provided in this embodiment.

[0143] Based on the video uploaded by the user, the above step 601 can be implemented as follows:

[0144] Extracting voice information and at least one frame of image containing an object from the video;

[0145] Performing classification and recognition processing on the at least one frame of image to determine the category corresponding to the object as a data category;

[0146] Convert voice information into text information and extract data search information from text information.

[0147] For at least one frame of image containing an object, the category of the object contained in each frame of image can be identified through various existing image classification schemes. If the object categories identified in all image frames are the same or mostly the same, the object category is used as the data category.

[0148] Extracting data search information from the above text information may specifically include: extracting keywords from the text information and using the extracted keywords as data search information. In the above example of "I want to buy a large screen and light weight", the extracted keywords are: large screen, light weight.

[0149] In this embodiment, the execution process of steps 602-603 can refer to the relevant descriptions in the aforementioned other embodiments and will not be repeated here.

[0150] The data search device of one or more embodiments of the present invention will be described in detail below. Those skilled in the art will appreciate that these data search devices can be configured using commercially available hardware components through the steps taught in this solution.

[0151] Figure 7Schematic diagram of a data search device provided by an embodiment of the present invention, the data search device can be arranged in the electronic device mentioned above. Figure 7 As shown, the device includes: a receiving module 11, a processing module 12, and an output module 13.

[0152] The receiving module 11 is used to receive data search information corresponding to a data category input by a user.

[0153] The processing module 12 is used to determine the attribute description information matching the data search information from the data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question and answer library corresponding to the data category; and to filter out the data set corresponding to the attribute value filtering condition from the candidate data sets corresponding to the data category according to the attribute value filtering condition corresponding to the attribute description information.

[0154] The output module 13 is used to output the data set.

[0155] Optionally, the data search information includes keywords corresponding to data attributes, and the data search information does not include data attribute values.

[0156] Optionally, the device further includes: an acquisition module and a screening module.

[0157] The acquisition module is used to acquire the characteristic information of the user.

[0158] The screening module is used to screen out a data set that matches the characteristic information of the user from the data set corresponding to the attribute value screening condition.

[0159] Optionally, the user's characteristic information includes at least one of the following: location information, historical search behavior.

[0160] Optionally, the processing module 12 is specifically used to: if the data search information does not exist in the data attribute library, encode the data search information to obtain a coding vector corresponding to the data search information; determine the similarity between the coding vectors corresponding to each attribute description information in the data attribute library and the coding vector corresponding to the data search information; determine that the attribute description information whose similarity with the coding vector corresponding to the data search information is greater than a set threshold is the attribute description information that matches the data search information.

[0161] Optionally, the device includes: a generating module.

[0162] The generation module is used to obtain multiple search corpora and multiple data attributes corresponding to the data category; determine the attribute description information corresponding to each of the multiple data attributes based on the multiple search corpora; obtain the attribute value screening conditions corresponding to each of the multiple data attributes; and generate the data attribute library based on the multiple data attributes, the attribute description information corresponding to each of the multiple data attributes, and the attribute value screening conditions.

[0163] Optionally, the generation module is specifically used to: determine the similarity between the encoding vector corresponding to each of the multiple search corpora and the encoding vector corresponding to each of the multiple data attributes; for any data attribute among the multiple data attributes, determine that the attribute description information corresponding to any data attribute includes a search corpus whose similarity with the encoding vector of any data attribute is greater than a set threshold.

[0164] Optionally, the generation module is also used to: perform synonym expansion on the search corpus whose similarity with the encoding vector of any data attribute is greater than a set threshold; and determine that the attribute description information corresponding to any data attribute includes information obtained after the synonym expansion.

[0165] Optionally, the generation module is also used to: perform clustering processing on the attribute description information corresponding to any data attribute; if multiple categories of attribute description information corresponding to any data attribute are obtained through the clustering processing, then obtain attribute value filtering conditions corresponding to the multiple categories of attribute description information respectively.

[0166] Optionally, the generating module is specifically used to: obtain a search corpus corresponding to the data category; perform word segmentation processing on each search corpus contained in the search corpus;

[0167] A preset number of target words with the highest word frequency are determined; and a search corpus containing any target word among the target words is selected from the search corpus set as the multiple search corpora.

[0168] Optionally, the output module 13 is further used to determine the display order of each data in the data set according to the attribute value corresponding to each data in the data set.

[0169] Optionally, the output module 13 is further used to: detect abnormal attribute values ​​in the data set; remove data corresponding to the abnormal attribute values, or arrange the data corresponding to the abnormal attribute values ​​after the data corresponding to non-abnormal attribute values.

[0170] Optionally, the data category is a commodity category, and the data set is a commodity set.

[0171] Optionally, the question-answering database is a knowledge base, and the knowledge base includes entities and entity relationships corresponding to the data categories. The processing module 12 can be specifically used to: if the data search information does not include entities in the knowledge base, determine that there is no reference query information matching the data search information in the knowledge base.

[0172] Figure 7 The device shown can perform the aforementioned Figures 1 to 4 The data search method provided in the illustrated embodiment, the detailed execution process and technical effects can be found in the description of the aforementioned embodiment, which will not be repeated here.

[0173] In one possible design, the above Figure 7 The structure of the data search device shown can be implemented as an electronic device, such as Figure 8 As shown, the electronic device may include: a first processor 21 and a first memory 22. The first memory 22 stores executable code. When the executable code is executed by the first processor 21, the first processor 21 can at least implement the above-mentioned Figures 1 to 4 The data search method provided in the illustrated embodiment.

[0174] Optionally, the electronic device may further include a first communication interface 23 for communicating with other devices.

[0175] In addition, an embodiment of the present invention provides a non-transitory machine-readable storage medium, wherein the non-transitory machine-readable storage medium stores executable code, and when the executable code is executed by a processor of an electronic device, the processor can at least implement the above-mentioned Figures 1 to 4 The data search method provided in the illustrated embodiment.

[0176] Fig. 9 FIG. 1 is a schematic diagram of the structure of another data search device provided by an embodiment of the present invention, and the data search device can be arranged in the electronic device described above. Fig. 9 As shown, the device includes: an identification module 31, an acquisition module 32, and an output module 33.

[0177] The recognition module 31 is used to recognize data search information corresponding to a data category from an image in response to an image input by a user.

[0178] The acquisition module 32 is used to determine the attribute description information matching the data search information from the data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question and answer library corresponding to the data category; and to filter out the data set corresponding to the attribute value filtering condition from the candidate data sets corresponding to the data category according to the attribute value filtering condition corresponding to the attribute description information.

[0179] The output module 33 is used to output the data set.

[0180] Optionally, the recognition module 31 may be specifically used to: perform character recognition processing on the image to obtain text contained in the image; and extract keywords corresponding to data attributes from the text as the data search information.

[0181] Optionally, the image includes an object, and the recognition module 31 can be specifically used to: determine a category corresponding to the object as the data category.

[0182] Fig. 9 The device shown can perform the aforementioned Figure 5 The data search method provided in the illustrated embodiment, the detailed execution process and technical effects can be found in the description of the aforementioned embodiment, which will not be repeated here.

[0183] In one possible design, the above Fig. 9 The structure of the data search device shown can be implemented as an electronic device, such as Fig.10 As shown, the electronic device may include: a second processor 41 and a second memory 42. The second memory 42 stores executable code. When the executable code is executed by the second processor 41, the second processor 41 can at least implement the above-mentioned Figure 5 The data search method provided in the illustrated embodiment.

[0184] Optionally, the electronic device may further include a second communication interface 43 for communicating with other devices.

[0185] In addition, an embodiment of the present invention provides a non-transitory machine-readable storage medium, wherein the non-transitory machine-readable storage medium stores executable code, and when the executable code is executed by a processor of an electronic device, the processor can at least implement the above-mentioned Figure 5 The data search method provided in the illustrated embodiment.

[0186] Fig.11 FIG. 1 is a schematic diagram of the structure of another data search device provided by an embodiment of the present invention, and the data search device can be arranged in the electronic device described above. Fig.11 As shown, the device includes: an identification module 51, an acquisition module 52, and an output module 53.

[0187] The identification module 51 is used to identify data search information corresponding to the data category from the video in response to the video input by the user.

[0188] The acquisition module 52 is used to determine the attribute description information matching the data search information from the data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question and answer library corresponding to the data category; and to filter out the data set corresponding to the attribute value filtering condition from the candidate data sets corresponding to the data category according to the attribute value filtering condition corresponding to the attribute description information.

[0189] The output module 53 is used to output the data set.

[0190] Optionally, the video includes an object, and the recognition module 51 can be specifically used to: extract voice information and at least one frame of image containing the object from the video; perform classification and recognition processing on the at least one frame of image to determine the category corresponding to the object as the data category; convert the voice information into text information; and extract the data search information from the text information.

[0191] Fig.11 The device shown can perform the aforementioned Figure 6 The data search method provided in the illustrated embodiment, the detailed execution process and technical effects can be found in the description of the aforementioned embodiment, which will not be repeated here.

[0192] In one possible design, the above Fig.11 The structure of the data search device shown can be implemented as an electronic device, such as Fig.12 As shown, the electronic device may include: a third processor 61 and a third memory 62. The third memory 62 stores executable code. When the executable code is executed by the third processor 61, the third processor 61 can at least implement the above-mentioned Figure 6 The data search method provided in the illustrated embodiment.

[0193] Optionally, the electronic device may further include a third communication interface 63 for communicating with other devices.

[0194] In addition, an embodiment of the present invention provides a non-transitory machine-readable storage medium, wherein the non-transitory machine-readable storage medium stores executable code, and when the executable code is executed by a processor of an electronic device, the processor can at least implement the above-mentioned Figure 6 The data search method provided in the illustrated embodiment.

[0195] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. Those of ordinary skill in the art may understand and implement the present invention without creative effort.

[0196] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by adding a necessary general hardware platform, and of course can also be implemented by combining hardware and software. Based on such an understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a computer product, and the present invention can be in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0197] The data search method provided in the embodiment of the present invention can be executed by a certain program / software, and the program / software can be provided by the network side. The electronic device mentioned in the above embodiment can download the program / software to a local non-volatile storage medium, and when it needs to execute the above data search method, the program / software is read into the memory by the CPU, and then the CPU executes the program / software to implement the data search method provided in the above embodiment. The execution process can refer to the above Figures 1 to 6 Instructions in .

[0198] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data search method, characterized in that: include: receiving data search information corresponding to a data category input by a user, wherein the data search information includes keywords corresponding to data attributes, and the data search information does not include data attribute values; If there is no reference query information matching the data search information in the question-answer library corresponding to the data category, determining attribute description information matching the data search information from a data attribute library corresponding to the data category, wherein the data attribute library at least includes a plurality of data attributes and attribute description information corresponding to each of the plurality of data attributes; According to the attribute value screening condition corresponding to the attribute description information, screening out a data set corresponding to the attribute value screening condition from the candidate data sets corresponding to the data category; outputting the data set; The method of determining the attribute description information matching the data search information from the data attribute library corresponding to the data category includes: if the data search information does not exist in the data attribute library, encoding the data search information to obtain a coding vector corresponding to the data search information; determining the similarity between the coding vectors corresponding to each attribute description information in the data attribute library and the coding vector corresponding to the data search information; and determining that the attribute description information whose similarity with the coding vectors corresponding to the data search information is greater than a set threshold is the attribute description information matching the data search information.

2. The method according to claim 1, characterized in that The method further comprises: Acquiring characteristic information of the user; A data set matching the characteristic information of the user is screened out from the data sets corresponding to the attribute value screening condition.

3. The method according to claim 2, characterized in that The user's characteristic information includes at least one of the following: Location information, historical search behavior.

4. The method according to any one of claims 1 to 3, characterized in that The method further comprises: Acquire multiple search corpora and multiple data attributes corresponding to the data category; Determining attribute description information corresponding to each of the plurality of data attributes according to the plurality of search corpora; Obtaining attribute value screening conditions corresponding to each of the plurality of data attributes; The data attribute library is generated according to the multiple data attributes, the attribute description information corresponding to each of the multiple data attributes, and the attribute value screening condition.

5. The method according to claim 4, characterized in that The determining, according to the plurality of search corpora, the attribute description information corresponding to each of the plurality of data attributes comprises: Determine the similarity between the encoding vectors corresponding to each of the plurality of search corpora and the encoding vectors corresponding to each of the plurality of data attributes; For any data attribute among the multiple data attributes, it is determined that the attribute description information corresponding to the any data attribute includes a search corpus whose similarity with the encoding vector of the any data attribute is greater than a set threshold.

6. The method according to claim 5, characterized in that The method further comprises: Performing synonym expansion on the search corpus whose similarity with the encoding vector of any data attribute is greater than a set threshold; It is determined that the attribute description information corresponding to any one of the data attributes includes information obtained after the synonym expansion.

7. The method according to claim 5, characterized in that The method further comprises: Performing clustering processing on the attribute description information corresponding to any of the data attributes; If multiple categories of attribute description information corresponding to any one of the data attributes are obtained through the clustering process, then the attribute value screening conditions corresponding to the multiple categories of attribute description information are obtained.

8. The method according to claim 4, characterized in that The obtaining of a plurality of search corpora corresponding to the data category includes: Obtaining a search corpus corresponding to the data category; Performing word segmentation processing on each search corpus included in the search corpus set; Determine a preset number of target words with the highest word frequency; A search corpus containing any target word among the multiple target words is selected from the search corpus set as the multiple search corpuses.

9. The method according to claim 1, characterized in that: The method further comprises: The display order of each data in the data set is determined according to the attribute value corresponding to each data in the data set.

10. The method according to claim 9, characterized in that The determining the display order of each data in the data set includes: Detecting abnormal attribute values ​​present in the data set; The data corresponding to the abnormal attribute value is eliminated, or the data corresponding to the abnormal attribute value is arranged after the data corresponding to the non-abnormal attribute value.

11. The method according to claim 1, characterized in that: The data category is a commodity category, and the data set is a commodity set.

12. The method according to claim 1, characterized in that The question-and-answer database is a knowledge base, which includes entities and entity relationships corresponding to the data categories; The method further comprises: If the data search information does not include an entity in the knowledge base, it is determined that there is no reference query information matching the data search information in the knowledge base.

13. A data search device, characterized in that: include: A receiving module, configured to receive data search information corresponding to a data category input by a user, wherein the data search information includes keywords corresponding to data attributes, and does not include data attribute values; a processing module, configured to determine attribute description information matching the data search information from a data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question-answer library corresponding to the data category; and to filter out a data set corresponding to the attribute value filtering condition from candidate data sets corresponding to the data category according to an attribute value filtering condition corresponding to the attribute description information, wherein the data attribute library contains at least a plurality of data attributes and attribute description information corresponding to each of the plurality of data attributes; An output module, used for outputting the data set; The processing module is specifically used to encode the data search information if the data search information does not exist in the data attribute library to obtain a coding vector corresponding to the data search information; determine the similarity between the coding vectors corresponding to each attribute description information in the data attribute library and the coding vector corresponding to the data search information; and determine that the attribute description information whose similarity with the coding vector corresponding to the data search information is greater than a set threshold is the attribute description information that matches the data search information.

14. An electronic device, characterized in that: include: A memory and a processor; wherein the memory stores executable code, and when the executable code is executed by the processor, the processor executes the data search method as described in any one of claims 1 to 12.

15. A non-transitory machine-readable storage medium, characterized in that: The non-transitory machine-readable storage medium stores executable code, and when the executable code is executed by a processor of an electronic device, the processor is caused to execute the data search method according to any one of claims 1 to 12.

16. A data search method, characterized in that: include: In response to an image input by a user, identifying data search information corresponding to a data category from the image, wherein the data search information includes a keyword corresponding to a data attribute, and the data search information does not include a data attribute value; If there is no reference query information matching the data search information in the question-answer library corresponding to the data category, determining attribute description information matching the data search information from a data attribute library corresponding to the data category, wherein the data attribute library at least includes a plurality of data attributes and attribute description information corresponding to each of the plurality of data attributes; According to the attribute value screening condition corresponding to the attribute description information, screening out a data set corresponding to the attribute value screening condition from the candidate data sets corresponding to the data category; outputting the data set; The method of determining the attribute description information matching the data search information from the data attribute library corresponding to the data category includes: if the data search information does not exist in the data attribute library, encoding the data search information to obtain a coding vector corresponding to the data search information; determining the similarity between the coding vectors corresponding to each attribute description information in the data attribute library and the coding vector corresponding to the data search information; and determining that the attribute description information whose similarity with the coding vectors corresponding to the data search information is greater than a set threshold is the attribute description information matching the data search information.

17. The method according to claim 16, characterized in that The step of identifying data search information corresponding to a data category from the image includes: Performing character recognition processing on the image to obtain text contained in the image; Keywords corresponding to data attributes are extracted from the text as the data search information.

18. The method according to claim 16, characterized in that The image includes an object, and the method further includes: Determine the category corresponding to the item as the data category.

19. A data search device, characterized in that: include: an identification module, configured to identify data search information corresponding to a data category from an image input by a user, wherein the data search information includes keywords corresponding to data attributes and does not include data attribute values; an acquisition module, configured to determine attribute description information matching the data search information from a data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question-answer library corresponding to the data category; and to filter out a data set corresponding to the attribute value filtering condition from candidate data sets corresponding to the data category according to an attribute value filtering condition corresponding to the attribute description information, wherein the data attribute library contains at least a plurality of data attributes and attribute description information corresponding to each of the plurality of data attributes; An output module, used for outputting the data set; The acquisition module is specifically used to encode the data search information if the data search information does not exist in the data attribute library to obtain a coding vector corresponding to the data search information; determine the similarity between the coding vectors corresponding to each attribute description information in the data attribute library and the coding vector corresponding to the data search information; and determine that the attribute description information whose similarity with the coding vector corresponding to the data search information is greater than a set threshold is the attribute description information that matches the data search information.

20. An electronic device, characterized in that: include: A memory and a processor; wherein the memory stores executable code, and when the executable code is executed by the processor, the processor executes the data search method as described in any one of claims 16 to 18.

21. A non-transitory machine-readable storage medium, characterized in that The non-transitory machine-readable storage medium stores executable code, and when the executable code is executed by a processor of an electronic device, the processor is caused to execute the data search method according to any one of claims 16 to 18.

22. A data search method, characterized in that: include: In response to a video input by a user, identifying data search information corresponding to a data category from the video, wherein the data search information includes a keyword corresponding to a data attribute, and the data search information does not include a data attribute value; If there is no reference query information matching the data search information in the question-answer library corresponding to the data category, determining attribute description information matching the data search information from a data attribute library corresponding to the data category, wherein the data attribute library at least includes a plurality of data attributes and attribute description information corresponding to each of the plurality of data attributes; According to the attribute value screening condition corresponding to the attribute description information, screening out a data set corresponding to the attribute value screening condition from the candidate data sets corresponding to the data category; outputting the data set; The method of determining the attribute description information matching the data search information from the data attribute library corresponding to the data category includes: if the data search information does not exist in the data attribute library, encoding the data search information to obtain a coding vector corresponding to the data search information; determining the similarity between the coding vectors corresponding to each attribute description information in the data attribute library and the coding vector corresponding to the data search information; and determining that the attribute description information whose similarity with the coding vectors corresponding to the data search information is greater than a set threshold is the attribute description information matching the data search information.

23. The method according to claim 22, characterized in that The video includes an object, and identifying data search information corresponding to a data category from the video includes: Extracting voice information and at least one frame of image containing the object from the video; Performing classification and recognition processing on the at least one frame of image to determine the category corresponding to the object as the data category; Converting the voice information into text information; The data search information is extracted from the text information.

24. A data search device, characterized in that: include: an identification module, configured to identify data search information corresponding to a data category from a video input by a user, wherein the data search information includes keywords corresponding to data attributes and does not include data attribute values; an acquisition module, configured to determine attribute description information matching the data search information from a data attribute library corresponding to the data category if there is no reference query information matching the data search information in the question-answer library corresponding to the data category; and to filter out a data set corresponding to the attribute value filtering condition from candidate data sets corresponding to the data category according to an attribute value filtering condition corresponding to the attribute description information, wherein the data attribute library contains at least a plurality of data attributes and attribute description information corresponding to each of the plurality of data attributes; An output module, used for outputting the data set; The acquisition module is specifically used to encode the data search information if the data search information does not exist in the data attribute library to obtain a coding vector corresponding to the data search information; determine the similarity between the coding vectors corresponding to each attribute description information in the data attribute library and the coding vector corresponding to the data search information; and determine that the attribute description information whose similarity with the coding vector corresponding to the data search information is greater than a set threshold is the attribute description information that matches the data search information.

25. An electronic device, characterized in that: include: A memory and a processor; wherein the memory stores executable code, and when the executable code is executed by the processor, the processor executes the data search method as described in claim 22 or 23.

26. A non-transitory machine-readable storage medium, characterized in that The non-transitory machine-readable storage medium stores executable code, and when the executable code is executed by a processor of an electronic device, the processor executes the data search method as described in claim 22 or 23.

Images