Information search method, device, electronic device and storage medium

Abstract

The embodiments of the present application disclose an information search method, device, electronic device, and storage medium. The method includes: based on the similarity between the search text and multiple tags, obtaining a reference tag set and similarity parameters corresponding to the reference tag set from multiple tags, the similarity parameters including the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameters are sorted in order of size; obtaining a reference information set corresponding to the reference tag set based on the correspondence between the tags and the information; determining the positional feature parameters of each information in the reference information set based on the sorted position of the tag corresponding to each information in the reference information set in the reference tag set; and obtaining the information corresponding to the search text from the reference information set based on the similarity parameters and the positional feature parameters of each information. This method makes the found information more relevant to the search text, thereby improving the accuracy of information search.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and more specifically, to an information search method, device, electronic device, and storage medium. Background Art

[0002] When creating a video or rendering, you may need to collect some information, such as some source images needed for the video or rendering. This information can be found by searching in an information search system. However, the accuracy of related information search systems still needs to be improved. Summary of the Invention

[0003] In view of the above problems, the present application proposes an information search method, device, electronic device and storage medium to improve the above problems.

[0004] In a first aspect, the present application provides an information search method, the method comprising: based on the similarity between a search text and a plurality of tags, obtaining a reference tag set and a similarity parameter corresponding to the reference tag set from the plurality of tags, the similarity parameter including the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameter being sorted in order of size; obtaining a reference information set corresponding to the reference tag set according to the correspondence between tags and information; determining a positional feature parameter of each information based on the sorting position of the tag corresponding to each information in the reference information set in the reference tag set, the sorting position of the tag in the reference tag set corresponding to the sorting position of the corresponding similarity in the similarity matrix; obtaining the information corresponding to the search text from the reference information set based on the similarity parameter and the positional feature parameter of each information.

[0005] In a second aspect, the present application provides an information search device, which includes: a similarity calculation unit, an information acquisition unit, a position feature acquisition unit, and an information screening unit. The similarity calculation unit is used to obtain a reference tag set and a similarity parameter corresponding to the reference tag set from the multiple tags based on the similarity between the search text and the multiple tags, the similarity parameter including the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameter are sorted in order of size; the information acquisition unit is used to obtain a reference information set corresponding to the reference tag set based on the correspondence between the tag and the information; the position feature acquisition unit is used to determine the position feature parameter of each information in the reference information set based on the sorting position of the tag corresponding to each information in the reference information set in the reference tag set, the sorting position of the tag in the reference tag set corresponding to the sorting position of the corresponding similarity in the similarity matrix; the information screening unit is used to obtain the information corresponding to the search text from the reference information set based on the similarity parameter and the position feature parameter of each information.

[0006] In a third aspect, the present application provides an electronic device comprising a processor and a memory; one or more programs are stored in the memory and configured to be executed by the processor to implement the above method.

[0007] In a fourth aspect, the present application provides a computer-readable storage medium, in which program code is stored, wherein the above method is executed when the program code is executed by a processor.

[0008] The present application provides an information search method, device, electronic device and storage medium. After obtaining the search text, the method can obtain a reference tag set from the multiple tags based on the similarity between the search text and multiple tags, as well as similarity parameters corresponding to the reference tag set. The similarity parameters include the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameters are sorted in order of size. Then, based on the correspondence between the tags and the information, the reference information set corresponding to the reference tag set is obtained, and based on the sorting position of the tag corresponding to each information in the reference information set in the reference tag set, the position feature parameter of each information is determined. The sorting position of the tag in the reference tag set corresponds to the sorting position of the corresponding similarity in the similarity matrix. Based on the similarity parameters and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0009] Therefore, through the above method, a reference information set can be obtained first through the similarity between the label corresponding to the information and the search text, and then after obtaining the position feature parameters corresponding to each information and the similarity ranking, the information corresponding to the search text can be filtered out from the reference information set based on the position feature parameters of each information and the similarity parameters corresponding to the reference information set, so that the found information is more relevant to the search text, so as to improve the accuracy of information search. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application, 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 application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A schematic diagram showing an application environment involved in an embodiment of the present application is shown;

[0012] Figure 2 A flowchart of an information search method proposed in one embodiment of the present application is shown;

[0013] Figure 3 A physical diagram of determining position feature parameters proposed in another embodiment of the present application is shown;

[0014] Figure 4 Another entity diagram for determining position feature parameters proposed in another embodiment of the present application is shown;

[0015] Figure 5 A flowchart of an information search method proposed in another embodiment of the present application is shown;

[0016] Figure 6 A schematic diagram showing that each range selection option proposed in this application corresponds to a storage area;

[0017] Figure 7 Shown Figure 5 Flowchart of the steps included in S250;

[0018] Figure 8 A flowchart of an information search method proposed in another embodiment of the present application is shown;

[0019] Figure 9 A schematic diagram of obtaining a vector corresponding to a label proposed in this application is shown;

[0020] Figure 10 A flowchart of an information search method proposed in another embodiment of the present application is shown;

[0021] Figure 11 A schematic diagram showing a user operation interface in an embodiment of the present application is shown;

[0022] Figure 12 A schematic diagram showing a user operation interface in a game platform according to an embodiment of the present application is shown;

[0023] Figure 13 A schematic diagram showing a user operation interface in a video platform according to an embodiment of the present application is shown;

[0024] Figure 14 A structural block diagram of a text processing device proposed in an embodiment of the present application is shown;

[0025] Figure 15 A structural block diagram of a text processing device proposed in another embodiment of the present application is shown;

[0026] Figure 16 A structural block diagram of a text processing device proposed in another embodiment of the present application is shown;

[0027] Figure 17 A structural block diagram of an electronic device for executing the information search method according to an embodiment of the present application is shown;

[0028] Figure 18 A storage unit for storing or carrying program codes for implementing the information search method according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0030] Artificial Intelligence (AI) refers to the theories, methods, techniques, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that seeks to understand the essence of intelligence and produce new intelligent machines that can respond in a manner similar to human intelligence. AI also involves studying the design principles and implementation methods of various intelligent machines, enabling them to possess the capabilities of perception, reasoning, and decision-making.

[0031] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, speech processing, computing, and machine learning / deep learning.

[0032] With the development of text processing technology within artificial intelligence, numerous scenarios involving text processing based on text processing and computer technology have emerged. For example, information search scenarios exist. For example, if a user needs to collect information while creating a video or rendering, they can use an information search system to search for information, such as source images. However, after researching related information search methods, the inventors discovered that the accuracy of related information search systems still needs to be improved.

[0033] Therefore, in order to improve the above-mentioned problem, the inventors have proposed an information search method, device, electronic device and storage medium. After obtaining the search text, a reference tag set and a similarity parameter corresponding to the reference tag set can be obtained from the multiple tags based on the similarity between the search text and multiple tags. The similarity parameter includes the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameter are sorted in order of size. Then, based on the correspondence between the tag and the information, a reference information set corresponding to the reference tag set is obtained, and based on the sorting position of the tag corresponding to each information in the reference information set in the reference tag set, the position feature parameter of each information is determined. The sorting position of the tag in the reference tag set corresponds to the sorting position of the corresponding similarity in the similarity matrix. Then, based on the similarity parameter and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0034] Therefore, through the above method, a reference information set can be obtained first through the similarity between the label corresponding to the information and the search text, and then after obtaining the position feature parameters corresponding to each information and the similarity ranking, the information corresponding to the search text can be filtered out from the reference information set based on the position feature parameters of each information and the similarity parameters corresponding to the reference information set, so that the found information is more relevant to the search text, so as to improve the accuracy of information search.

[0035] Before further describing the embodiments of the present application in detail, an application environment involved in the embodiments of the present application is introduced.

[0036] like Figure 1 As shown, Figure 1 The following is a schematic diagram of an application environment according to an embodiment of the present application. The application environment includes a client 110 and a server 120. The client 110 is configured to collect text input by a user and then send the collected text to the server 120. After receiving the text input by the user, the server 120 may perform an information search based on the information search method provided in the present application and return the searched information corresponding to the text input by the user to the client 110. The client 110 then displays the returned information.

[0037] It should be noted that Figure 1 This is an exemplary application environment, and the method provided in the embodiment of the present application can also be run in other application environments. For example, the information search method provided in the embodiment of the present application can be executed by the client 110.

[0038] It should be noted that the server 120 may be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. The electronic device where the client 110 is located may be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited thereto.

[0039] The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0040] See also Figure 2 , Figure 2 FIG2 is a flow chart of an information search method proposed in an embodiment of the present application, the method comprising:

[0041] S110: Based on the similarity between the search text and multiple tags, a reference tag set and a similarity parameter corresponding to the reference tag set are obtained from the multiple tags, where the similarity parameter includes the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameter are sorted in order of size.

[0042] It should be noted that in the embodiments of the present application, all pre-stored information may correspond to labels. The stored information may be a video, a picture, or text, etc. The label corresponding to the information may be used to identify the content features corresponding to the information. For example, taking the information as a picture, if the content of the picture is a female warrior holding a feather fan, then the labels corresponding to the picture may include feather fan and female warrior. For another example, if the content of the picture is a football player playing football, then the labels corresponding to the picture may include football, playing football, and football player.

[0043] If the information corresponds to a tag, a reference tag set can be determined from the multiple tags by comparing the similarity between the search text and the multiple tags. The search text can be understood as the text based on which the information search is performed, or it can be understood as the information being searched being information related to the search text. The multiple tags are used to calculate similarity with the search text.

[0044] In this embodiment, the reference tag set can be understood as a set of tags whose similarities with the search text meet specified conditions.

[0045] As a way, the specified condition can be a plurality of tags whose corresponding similarity rankings are at the top of the specified digits. For example, if the plurality of tags include tag t1, tag t2, tag t3, tag t4 and tag t5. Among them, if the similarity between tag t1 and the search text is s1, if the similarity between tag t2 and the search text is s2, if the similarity between tag t3 and the search text is s3, if the similarity between tag t4 and the search text is s4, if the similarity between tag t5 and the search text is s5, wherein s1 is greater than s2, s2 is greater than s3, s3 is greater than s4, and s4 is greater than s5, then the plurality of similarities are sorted from large to small as s1, s2, s3, s4 and s5, wherein the larger the similarity value, the higher the ranking. Then if the number of the top specified digits is 3, then it can be determined that the tag t1 corresponding to similarity s1, the tag t2 corresponding to similarity s2 and the tag t3 corresponding to similarity s3 are tags in the reference tag set.

[0046] It should be noted that, in addition to being 3, the number of digits specified for the top of the sort can also be dynamically adjusted according to actual needs. For example, if the number of digits specified for the top of the sort is greater, the number of tags in the reference tag set may be greater, and the amount of calculation involved in subsequent steps will be greater, which may reduce the rate of finding information. As a way, the number of digits specified for the top of the sort can be dynamically configured according to the current information search requirements. If the current information search requirement is quantity priority, the number of digits specified for the top of the sort can be configured as the first digit. If the current information search requirement is rate priority, the number of digits specified for the top of the sort can be configured as the second digit, wherein the second digit is smaller than the first digit.

[0047] As another way, the specified condition can be a tag whose corresponding similarity is greater than a specified similarity threshold. Taking the aforementioned multiple tags including tag t1, tag t2, tag t3, tag t4 and tag t5 as an example, if the similarity between tag t1 and the search text is s1, if the similarity between tag t2 and the search text is s2, if the similarity between tag t3 and the search text is s3, if the similarity between tag t4 and the search text is s4, if the similarity between tag t5 and the search text is s5. If the specified similarity threshold is s6, and s1 and s2 are greater than s6, and s3, s4 and s5 are less than s6, then it can be determined that tag t1 corresponding to similarity s1 and tag t2 corresponding to s2 are tags in the reference tag set. Similarly, the specified similarity threshold can also be adjusted according to actual conditions. Optionally, if the current information search requirement is quantity-first, the specified similarity threshold can be configured as the first threshold. If the current information search requirement is rate-first, the specified similarity threshold can be configured as the second threshold, wherein the second threshold is smaller than the first threshold.

[0048] After obtaining the reference tag set, the similarities corresponding to the tags in the reference tag set can be combined into a similarity parameter, and the similarities in the similarity parameter are sorted in order of size. For example, if the reference tag set includes tags t1, t2, and t3, and the similarity between tag t1 and the search text is s1, the similarity between tag t2 and the search text is s2, and the similarity between tag t3 and the search text is s3, then the similarities included in the similarity parameter corresponding to the reference tag set are similarity s1, similarity s2, and similarity s3. And when similarity s1 is greater than s2, and similarity s2 is greater than similarity s3, in the similarity parameter, similarity s1 is ranked first, similarity s2 is ranked after similarity s1, and similarity s3 is ranked after similarity s2.

[0049] S120: Acquire a reference information set corresponding to the reference tag set according to the correspondence between the tag and the information.

[0050] In this embodiment, each stored information can have its own corresponding label, where the correspondence between the label and the information represents which labels each stored information corresponds to. After obtaining a reference label set, the information corresponding to each label in the reference label set can be obtained based on the correspondence between the label and the information, and the information corresponding to each label can be combined into a reference information set. For example, if the reference label set includes label t1, label t2, and label t3, where label t1 corresponds to information i1, label t2 corresponds to information i2, and label t3 corresponds to information i3, then the reference information set corresponding to the reference label set including label t1, label t2, and label t3 includes information i1, information i2, and information i3.

[0051] As a way to establish a multi-valued function between the label and the information as the corresponding relationship between the label and the information, the multi-valued function can be understood as an index, and then after obtaining the label, the information corresponding to the label can be obtained through the multi-valued function.

[0052] It should be noted that in the correspondence between tags and information, a piece of information may correspond to multiple tags, and thus different tags in the reference tag set may correspond to the same information. For example, if the reference tag set includes tags t1, t4, and t5, where tag t1 corresponds to information i1, tag t4 corresponds to information i4, and tag t5 corresponds to information i4, then the reference information set corresponding to the reference tag set including tags t1, t4, and t5 includes information i1 and information i4.

[0053] S130: Determine the position characteristic parameter of each information based on the ranking position of the label corresponding to each information in the reference information set in the reference label set, where the ranking position of the label in the reference label set corresponds to the ranking position of the corresponding similarity in the similarity matrix.

[0054] In this embodiment, after the reference information set is determined, the information corresponding to the search text will be screened out in the reference information set. In this embodiment, the similarities in the similarity parameters will be sorted according to the size of the similarities, and the sorting positions of the labels in the reference label set will also correspond to the sorting positions of the corresponding similarities in the similarity matrix. Among them, the sorting position correspondence can be understood as the sorting position of the label in the reference label set is the same as the sorting position of the similarity corresponding to the label in the similarity parameter. For example, if the sorting position of the similarity s1 corresponding to the label t1 in the similarity parameter is at the front, then the label t1 is also sorted at the front in the reference label set.

[0055] Then, when the labels in the reference label set are sorted according to the size of the similarity, the position feature parameters of each information are determined to correspond to the order of the similarity size according to the sort position of the label corresponding to each information in the reference label set, so that the position feature parameters corresponding to the information can represent the degree of similarity between the information and the search text.

[0056] As a method of determining the position characteristic parameters corresponding to information, the target information can be obtained first, and the target information is the information for which the position characteristic parameters are currently being determined. The characteristic value corresponding to the label corresponding to the target information in the reference label set can be assigned to a first value, and the characteristic values ​​corresponding to other labels in the reference label set can be assigned to a second value, and then the characteristic values ​​corresponding to each label in the reference label set are combined and sorted according to the sorting position of the label to obtain the position characteristic parameters corresponding to the information. Among them, the other labels are labels for the corresponding information that do not include the target information. In the case of multiple information, the above method of determining the position characteristic parameters is executed for each information to obtain the position characteristic parameters corresponding to each information.

[0057] For example, the first value is 1 and the second value is 0, such as Figure 3 As shown, the reference expression set includes label t1, label t2, label t3, label t4 and label t5, and the current target information is the information i1 corresponding to label t1, and there is no corresponding relationship between label t2, label t3, label t4 and label t5 and information i1. Then for the information i1, the feature value corresponding to label t1 is 1, and the feature values ​​corresponding to label t2, label t3, label t4 and label t5 are all 0. The resulting position feature parameter is "10000". Furthermore, if the current target information is that label t2, label t3, label t4 and label t5 all have corresponding information i2, and label t1 does not correspond to information i2, then the determined position feature parameter can be as follows Figure 4 The display shows "01111".

[0058] S140: Based on the similarity parameter and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0059] After obtaining the similarity parameter and the position feature parameter corresponding to each piece of information, the information corresponding to the search text can be obtained from the reference information set based on the similarity parameter and the position feature parameter of each piece of information.

[0060] Take the case where both the similarity parameter and the position feature parameter are in matrix form. In this case, the similarity parameter is a similarity matrix, and the position feature parameter is a position feature matrix. The matrix elements in the similarity matrix are the similarities in the aforementioned similarity parameters, and correspondingly, the matrix elements in the position feature matrix are the eigenvalues ​​included in the position feature parameters. For example, if the similarity parameters include similarity s1, similarity s2, similarity s3, similarity s4, and similarity s5. Then the similarity matrix in matrix form corresponding to the similarity parameter can be [s1, s2, s3, s4, s5]. Correspondingly, if the position feature parameter is "10000", then the position feature matrix in matrix form corresponding to the position feature parameter can be [1, 0, 0, 0, 0].

[0061] For each piece of information, the similarity matrix can be multiplied by the positional feature matrix of each piece of information to obtain the corresponding matching value for each piece of information. Based on the matching value, the information corresponding to the search text can be obtained. It can be understood that the matching value corresponding to each piece of information represents the degree of similarity between the information and the search text. The larger the corresponding matching value, the more relevant the information is to the search text.

[0062] As a method, in an embodiment of the present application, there may be an interface for configuring the number of information searches. The user can use this interface to configure the upper limit of the number of information searches during each search process to obtain the upper limit of the information search. After obtaining the matching value corresponding to each information, the matching value corresponding to each information can be sorted based on the size of the matching value, and the information whose matching value sorting order does not exceed the upper limit of the information search is used as the information corresponding to the search text. Among them, the sorting order does not exceed the upper limit of the information search can be understood as the order value corresponding to the sorting order is less than or equal to the upper limit of the information search. It is rational. The matching value corresponding to information i1 is m1, the matching value corresponding to information i2 is m2, and the matching value corresponding to information i3 is m3. Where m1 is greater than m2, and m2 is greater than m3, then the matching value corresponding to information i1 is sorted in order of 1, the matching value corresponding to information i2 is sorted in order of 2, and the matching value corresponding to information i3 is sorted in order of 3. If the upper limit of the information search is 2, the information i1 with a matching value sorting order of 1 and the information i2 with a matching value sorting order of 2 will be used as the information corresponding to the search text.

[0063] The present application provides an information search method. After obtaining the search text, a reference tag set and similarity parameters corresponding to the reference tag set can be obtained from multiple tags based on the similarity between the search text and multiple tags. The similarity parameters include the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameters are sorted in order of size. Then, based on the correspondence between the tags and the information, a reference information set corresponding to the reference tag set is obtained, and based on the sorting position of the tag corresponding to each information in the reference information set in the reference tag set, the position feature parameter of each information is determined, wherein the sorting position of the tag in the reference tag set corresponds to the sorting position of the corresponding similarity in the similarity matrix. Then, based on the similarity parameters and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0064] Therefore, through the above method, a reference information set can be obtained first through the similarity between the label corresponding to the information and the search text, and then after obtaining the position feature parameters corresponding to each information and the similarity ranking, the information corresponding to the search text can be filtered out from the reference information set based on the position feature parameters of each information and the similarity parameters corresponding to the reference information set, so that the found information is more relevant to the search text, so as to improve the accuracy of information search.

[0065] See also Figure 5 , Figure 5 FIG2 is a flow chart of an information search method proposed in an embodiment of the present application, the method comprising:

[0066] S210: Based on the similarity between the search text and multiple tags, a reference tag set and a similarity parameter corresponding to the reference tag set are obtained from the multiple tags, where the similarity parameter includes the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameter are sorted in order of size.

[0067] As described above, multiple tags are the basis for obtaining a reference tag set. In the embodiment of the present application, there are multiple ways to determine multiple tags for similarity calculation with the search text.

[0068] As a method, the multiple tags used for similarity calculation with the search text can be all the stored tags. In this way, after the search text is obtained, the similarity calculation between the search text and all the stored tags can be performed respectively, and then the reference tag set can be obtained based on the method proposed in the above embodiment. Exemplarily, if all the stored tags include N tags, then the similarity calculation between the search text and the N tags will be performed respectively to obtain the similarity between the search text and the N tags, and then the reference tag set can be obtained from the N tags based on the similarity of each tag.

[0069] As another approach, the multiple tags used to calculate similarity with the search text can be some of the stored tags. It should be noted that when the number of stored tags is relatively large, if the search text is calculated to have a similarity with all stored tags each time an information search is performed, the amount of calculation may be large, and the search results may not be fed back to the user quickly. In this approach, a search range selection interface can be provided to the user, and in the process of calculating similarity, the tags corresponding to the search range selected by the user can be used as the multiple tags for similarity calculation with the search text.

[0070] Optionally, the search range selection interface may include a high-popularity tag priority option. Among them, the high-popularity tag priority representation will filter tags from the more popular tags as a reference tag set, so that the information finally queried is the currently popular information. Then if the high-popularity tag priority option is detected, the corresponding tags whose search heat is greater than the specified heat threshold can be used as multiple tags for similarity calculation with the search text. In this way, a heat field can be configured for each tag when storing each tag, so as to characterize the heat corresponding to the tag based on the value of the heat field.

[0071] The popularity of a tag can be characterized based on the number of times the tag is used as a tag in a reference tag set. It should be noted that in the embodiment of the present application, the tags in the reference tag set are tags whose similarity with the search text meets a specified condition. Therefore, if the similarity between the tags in the reference tag set and the search text is higher than that of other tags, the popularity corresponding to the tags in the generated reference tag set can be increased by 1 each time a reference tag set is generated.

[0072] Optionally, the search range selection interface may include a low-heat tag priority option. Among them, the low-heat tag priority representation will filter tags from tags with lower heat as a reference tag set, so that the information finally queried is less used information, which is conducive to avoiding the duplication of videos or renderings produced using the searched information with those produced by others. Then, if the low-heat tag priority option is detected, the corresponding tags with a search heat lower than the specified heat threshold can be used as multiple tags for similarity calculation with the search text.

[0073] It should be noted that if the stored tags are all stored in the same storage area, then in the method of selecting multiple tags based on the search range selection interface, an operation of searching for tags corresponding to the search range selection interface will be added in advance, and in the process of searching for tags corresponding to the search range selection interface, all tags in the tag storage area will still be traversed, which will make the efficiency improvement of this method not obvious enough. In order to further improve the search efficiency brought about by the method of determining multiple tags through the search range selection interface, a storage area can be established for each range selection option in the search range selection interface, so that the tags corresponding to each range selection option can be stored in their respective corresponding storage areas in advance. In this case, when obtaining multiple tags for similarity calculation with the search text, the tags in the storage area corresponding to the selected range selection option can be directly used as multiple tags.

[0074] For example, Figure 6 As shown, the search range selection interface includes range selection option 1, range selection option 2, and range selection option 3. Among them, range selection option 1 corresponds to storage area 1, and the tags corresponding to range selection option 1 are stored in storage area 1. Range selection option 2 corresponds to storage area 2, and the tags corresponding to range selection option 2 are stored in storage area 2. Range selection option 3 corresponds to storage area 3, and the tags corresponding to range selection option 3 are stored in storage area 3. Then, when it is detected that the user has selected range selection option 1, the tags stored in storage area 1 will be used as multiple tags for similarity calculation with the search text.

[0075] Optionally, in this embodiment, the storage area can be understood as a data table or a folder.

[0076] S220: Acquire a reference information set corresponding to the reference tag set according to the correspondence between the tag and the information.

[0077] As one approach, in embodiments of the present application, the amount of information included in a reference information set can be limited, such that, during the process of determining the reference information set, the amount of information in the generated reference information set does not exceed the limit. Optionally, a reference information quantity configuration interface can be provided for user configuration. Through this reference information quantity configuration interface, the user can limit the amount of information in the reference information set generated during each information search process.

[0078] For example, if the reference tag set includes tag t1, tag t2, tag t3, tag t4, and tag t5, and each tag t1, tag t2, tag t3, tag t4, and tag t5 corresponds to a piece of information, and each tag corresponds to a different piece of information, then if the upper limit of the number of pieces of information in the reference information set is 5, the pieces of information corresponding to each of tags t1, tag t2, tag t3, tag t4, and tag t5 will be included in the reference information set. Correspondingly, if the number of pieces of information corresponding to the tags in the reference tag set exceeds the upper limit of the number of pieces of information in the reference information set, the information corresponding to some of the tags can be discarded based on the corresponding similarity. For example, if the upper limit of the number of pieces of information in the reference information set is 5, the information corresponding to tag t5, which has the lowest similarity, can be discarded. The resulting reference information set then includes the information corresponding to each of tags t1, tag t2, tag t3, and tag t4.

[0079] It should be noted that in some cases, the same tag may correspond to multiple pieces of information. Therefore, if there are multiple tags in the reference tag set, the amount of information corresponding to a particular tag may exceed the upper limit of the information in the reference information set, or the amount of information corresponding to some tags may exceed the upper limit of the information in the reference information set. In this case, the amount of information corresponding to each tag can be determined sequentially according to the order of the tags in the reference tag set to obtain information that does not exceed the upper limit of the information in the reference information set. During the process of obtaining the reference information set, the sum of the amount of information corresponding to the current tag and the amount of information corresponding to the preceding tag can be calculated to see whether it exceeds the upper limit of the information in the reference information set. If not, the above determination is repeated for the following tags. If it does, the information corresponding to the preceding tag and the supplementary information corresponding to the current tag are combined as information in the reference information set. The supplementary information can be randomly selected from the information corresponding to the current tag, and the amount of this supplementary information is the difference between the upper limit of the information in the reference information set and the amount of information corresponding to the preceding tag.

[0080] For example, if the reference tag set includes tags t1, t2, and t3, then the number of corresponding information will be determined for tag t1 first. In this case, tag t1 is the current tag. If it is determined that the number of information corresponding to tag t1 does not exceed the upper limit of the information in the reference information set, tag t2 will be further determined. In this case, tag t2 is the current tag. If the sum of the number of information corresponding to tag t2 and the number of information corresponding to tag t1 still does not exceed the upper limit of the information in the reference information set, tag t3 will be further determined. If the sum of the number of information corresponding to tag t3 and the number of information corresponding to tags t1 and t2 exceeds the upper limit of the information in the reference information set, the information corresponding to tags t1 and t2 and the supplementary information corresponding to tag t3 will be used as the information in the reference information set.

[0081] S230: Determine the position characteristic parameter of each information based on the ranking position of the label corresponding to each information in the reference information set in the reference label set, where the ranking position of the label in the reference label set corresponds to the ranking position of the corresponding similarity in the similarity matrix.

[0082] S240: Obtain a weight parameter corresponding to the reference tag set, where the weight parameter includes a weight corresponding to each tag in the reference tag set, wherein a tag with a greater similarity has a greater weight corresponding to it.

[0083] Each tag in the reference tag set can be assigned a weight. The weight is used to adjust the probability that the information corresponding to the tag is determined to be the search information. Accordingly, the larger the weight corresponding to the tag, the greater the probability that the information corresponding to the tag is determined to be the search information.

[0084] As a method for obtaining the weight parameters corresponding to the reference tag set, the weight parameters corresponding to each tag are configured with equal differences. In this method, the difference between the weights corresponding to each adjacent tag is the same, and the adjacent tags are the two tags with the closest similarity. For example, if the reference tag set includes tags t1, t2, and t3, then the difference between the weight corresponding to tag t1 and the weight corresponding to tag t2 is the same as the difference between the weight corresponding to tag t2 and the weight corresponding to tag t3.

[0085] As another approach, obtaining weight parameters corresponding to the reference tag set includes: determining a weight corresponding to each tag based on the similarity of the tags in the reference tag set to obtain the weight parameters; wherein the smaller the difference in similarity between adjacent tags in the reference tag set, the larger the difference in weights corresponding to the adjacent tags, where adjacent tags are the two tags with the closest similarity. Thus, when the similarity gap is small, the information corresponding to the tag with the greater similarity is more likely to be used as the information corresponding to the search text.

[0086] For example, if the reference tag set includes tag t1, tag t2, and tag t3, wherein the similarity s1 corresponding to tag t1 is greater than the similarity s2 corresponding to tag t2, and the similarity s2 corresponding to tag t2 is greater than the similarity s3 corresponding to tag t3, then in this case, the weight corresponding to tag t1 is greater than the weight corresponding to tag t2, and the weight corresponding to tag t2 is greater than the weight corresponding to tag t3. In addition, after determining the order of the weights corresponding to each tag, the difference in weights corresponding to each tag can be further determined. Optionally, a correspondence between the similarity difference and the weight difference can be established in advance, and the correspondence is inversely proportional, that is, the smaller the corresponding similarity difference, the larger the corresponding weight difference. Then, after determining the size relationship and difference relationship between the aforementioned weights, the weight of the tag with the smallest weight can be determined first, and then the weight corresponding to each tag can be determined through the size relationship and difference relationship.

[0087] For example, if the weight of the smallest tag t3 is w, and the difference between t3 and t2 is d1, then the weight of t2 can be determined to be w+d1. If the difference between t2 and t1 is d2, then the weight of t1 can be determined to be w+d1+d2.

[0088] S250: Based on the similarity parameter, the weight parameter and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0089] As a way, Figure 7 As shown, the similarity parameter is the similarity matrix, the position feature parameter is the position feature matrix, and the weight parameter is the weight matrix; based on the similarity parameter, the weight parameter, and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set, including:

[0090] S251: Obtain the product of the similarity matrix, the weight matrix, and the position feature matrix of each information as the matching value corresponding to each information.

[0091] S252: The information whose corresponding matching value satisfies the specified sorting condition is used as the information corresponding to the search text.

[0092] Optionally, the information with the largest matching value can be used as the information corresponding to the search text. In this way, the matching value of the information can be calculated based on the following formula, and the information corresponding to the search text can be determined based on the matching value. The formula is:

[0093]

[0094] Among them, H represents the weight matrix, S represents the similarity matrix, and L represents the position feature parameter matrix. Degree represents the matching value, and E represents the reference information set. The function represents the information with the largest corresponding matching value in the reference set.

[0095] The present application provides an information search method, which can first obtain a reference information set through the similarity between the label corresponding to the information and the search text through the above method, and then after obtaining the position feature parameters corresponding to each information and the similarity ranking, the information corresponding to the search text is filtered out from the reference information set based on the position feature parameters of each information and the similarity parameters corresponding to the reference information set, thereby making the found information more relevant to the search text to improve the accuracy of the information search. In addition, in this embodiment, a weight parameter including the weight corresponding to each label in the reference label set can also be obtained, and then the information corresponding to the search text is obtained from the reference information set through the similarity parameter, weight parameter and the position feature parameter of each information, so that the information search result can be adjusted by the weight parameter, thereby improving the flexibility of adjusting the information search result.

[0096] See also Figure 8 , Figure 8FIG2 is a flow chart of an information search method proposed in an embodiment of the present application, the method comprising:

[0097] S310: Obtain the vector corresponding to the search text.

[0098] One method for obtaining a vector corresponding to a search text includes: obtaining a one-hot encoding corresponding to the search text; inputting the one-hot encoding into a target neural network, the target neural network including a hidden layer; and using weight parameters output by the hidden layer as the vector corresponding to the search text. Optionally, the target neural network can be generated by training a word2vec model.

[0099] It should be noted that, similarly, the process of calculating the vector corresponding to the label and the method of obtaining the vector corresponding to the search text can be the same. Figure 9 As shown, for label t, we can first obtain the corresponding one-hot encoding, and then use the one-hot encoding of label t as the input data of the output layer of the target neural network, that is, Figure 9 The input layer vector l in is then input into the hidden layer, and the weight parameter of the hidden layer output is used as the vector of the label t, that is, Figure 9 The label vector t in .

[0100] S320: Calculate the vector similarity between the vector corresponding to the search text and the vectors of the multiple tags, and use the vector similarity as the similarity between the search text and the multiple tags.

[0101] Optionally, vector similarity can be calculated based on the following formula:

[0102]

[0103] Among them, similarity represents the calculated vector similarity. m and w i Represents the two vectors involved in the calculation.

[0104] It should be noted that, in the embodiment of the present application, after obtaining the vectors corresponding to the search text and the label, the similarity between the search text and the label can be calculated by calculating the vector similarity based on the aforementioned method. This method can achieve a more accurate determination of the degree of similarity between the search text and the label. In addition, in the embodiment of the present application, the similarity between the search text and the label can also be obtained by other methods. For example, the similarity can be determined by determining the number of words that the search text and the label jointly include. In this way, the greater the number of words that the search text and the label jointly include, the greater the corresponding similarity.

[0105] S330: Based on the similarity between the search text and multiple tags, a reference tag set and a similarity parameter corresponding to the reference tag set are obtained from the multiple tags, where the similarity parameter includes the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameter are sorted in order of size.

[0106] S340: Acquire a reference information set corresponding to the reference tag set according to the correspondence between the tag and the information.

[0107] S350: Determine the position characteristic parameter of each information based on the ranking position of the label corresponding to each information in the reference information set in the reference label set, wherein the ranking position of the label in the reference label set corresponds to the ranking position of the corresponding similarity in the similarity matrix.

[0108] S360: Based on the similarity parameter and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0109] The present application provides an information search method, which can first obtain a reference information set through the similarity between the label corresponding to the information and the search text through the above method, and then after obtaining the position feature parameters corresponding to each information and the similarity ranking, the information corresponding to the search text is filtered out from the reference information set based on the position feature parameters of each information and the similarity parameters corresponding to the reference information set, thereby making the found information more relevant to the search text, thereby improving the accuracy of the information search. Moreover, in this embodiment, the reference label set can be obtained based on the similarity between the vector corresponding to the search text and the vector of the label corresponding to the information, which is conducive to improving the labels in the obtained reference label set to be labels that are more relevant to the search text, so as to improve the accuracy of the information finally found.

[0110] See also Figure 10 , Figure 10 FIG2 is a flow chart of an information search method proposed in an embodiment of the present application, the method comprising:

[0111] S410: Obtain input text.

[0112] The input text may be understood as text input by the user that represents the query content desired by the user.

[0113] S420: Segment the input text to obtain multiple search texts.

[0114] As a method, the information input text is segmented to obtain multiple search texts, including: obtaining the segmentation length, the segmentation length is used to limit the upper limit of the length of the search text obtained by segmentation; the input text is segmented based on the segmentation length to obtain multiple search texts.

[0115] As another method, the input text is segmented to obtain multiple search texts, including: segmenting the input text multiple times, obtaining the results of the multiple segmentations as multiple search texts, wherein the segmentation length corresponding to each segmentation is different, and the segmentation length is used to limit the upper limit of the length of the search text obtained by the segmentation.

[0116] It should be noted that the results of word segmentation of the input text may be different when the word segmentation length is different. For example, if the input text is "female warrior holding a feather fan", if the word segmentation length is 2, then the word segmentation results may include "feather fan", "female" and "warrior". If the word segmentation length is 4, then the word segmentation results may include "female warrior", and the vectors corresponding to "female warrior" and "warrior" will be different. Therefore, in the process of similarity calculation, based on the same label, the two words "female warrior" and "warrior" will have different similarities. Therefore, by performing multiple word segmentations on the input text with different word segmentation lengths, the multiple search texts obtained can include more content, which is conducive to more accurate search of the search information corresponding to the input text.

[0117] Moreover, by changing the segmentation length, the order of some contents in the input text can be retained, and the information corresponding to the input text can be found more accurately. For example, taking the aforementioned input text "female warrior holding a feather fan" as an example, if the segmentation length is 2, then the word "female warrior" will be divided into "female" and "warrior", and then information search will be performed for "female" and "warrior" respectively. If there are multiple segmentation lengths, the segmentation text with the content of "female warrior" can also be obtained, and then the information search will be performed separately for the search text "female warrior", which is conducive to more accurate finding of information with the content of female and warrior.

[0118] S430: Based on the similarity between the search text and multiple tags, a reference tag set and a similarity parameter corresponding to the reference tag set are obtained from the multiple tags, where the similarity parameter includes the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameter are sorted in order of size.

[0119] S440: Acquire a reference information set corresponding to the reference tag set according to the correspondence between the tag and the information.

[0120] S450: Determine the position characteristic parameter of each information based on the ranking position of the label corresponding to each information in the reference information set in the reference label set, wherein the ranking position of the label in the reference label set corresponds to the ranking position of the corresponding similarity in the similarity matrix.

[0121] S460: Based on the similarity parameter and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0122] S470: Outputting the search information corresponding to each of the plurality of search texts as the search information corresponding to the input text.

[0123] The following is an introduction to the scenarios involved in the embodiments of the present application by means of schematic diagrams.

[0124] like Figure 11 As shown, in Figure 11 The user interface shown includes an input control 10, a search trigger control 11, and an information display control 12. The user can enter text using the input control 10. After completing the text entry, the user can use the search trigger control 11 to trigger a search for information corresponding to the input text. During this process, the information corresponding to the searched text can be obtained using the methods described in the previous embodiments. Finally, after obtaining the information corresponding to the searched text, the found information can be displayed in the information display control 12. The user can also perform operations such as deleting or transferring the information in the information display control 12.

[0125] The user operation interface can be applied to different platforms. Optionally, the platform can be a platform related to games. In the platform related to games, the information to be searched can be game pictures, and the description information entered can be description information about the game pictures. For example, Figure 12 As shown, if "archer" is input into the input control 10, then based on the method of segmenting the description information described above, the search text obtained by processing "archer" may include archer, bow and arrow, and warrior. Then, the platform can respectively execute the information search method provided in the embodiment of the present application based on the three search texts of archer, bow and arrow, and warrior, and output the pictures found based on each search text to the information display control 12 for display.

[0126] In addition, in addition to the platform for games, the user operation interface can also be applied to other platforms, for example, the user operation interface can also be applied to the platform for videos. In this way, the information to be searched can be a video, and the description information entered can be the description information about the video. For example, Figure 13 As shown, if "Let's play football" is input into the input control 10, the search text obtained after processing the description information may include "playing football" and "football". Then, the information search method provided by the embodiment of the present application can be executed using "playing football" and "football" as the search text, and the cover image corresponding to the found video is output to the information display control 12 for display. In addition, a play control control 13 can be displayed in the cover image so that after a touch operation is detected on the play control control 13, the video corresponding to the touched play control 13 is played.

[0127] The present application provides an information search method, whereby, through the above-mentioned method, the input text obtained can be segmented to obtain multiple search texts, and then a reference information set is obtained by the similarity between the label corresponding to the information and the search text. After obtaining the position feature parameters corresponding to the similarity ranking for each information, the information corresponding to the search text is filtered out from the reference information set based on the position feature parameters of each information and the similarity parameters corresponding to the reference information set, thereby making the found information more relevant to the search text to improve the accuracy of the information search. Then, when the required information is found more accurately for each search text, the corresponding searched information can be obtained more accurately for the input text.

[0128] It should be noted that corresponding tags can be pre-established for the information being searched in the embodiments of the present application. As a method, based on the similarity between the search text and multiple tags, before obtaining a reference tag set from multiple tags, the method further includes: obtaining initial description information corresponding to the information;

[0129] Perform text processing on the initial description information to obtain processed information; and use the initial description information and the processed information as labels corresponding to the information. The text processing includes at least one of the following processing methods: performing word segmentation on the initial description information; and obtaining synonyms corresponding to the initial description information.

[0130] It is understandable that the labels corresponding to the information can be annotated manually, which may result in the annotated labels being incomplete. The content annotated manually can be understood as the aforementioned initial description information. This incompleteness can be understood as not annotating all the content in the information. For example, in a picture of a female warrior holding a feather fan, the corresponding annotated labels may only include "female" and "warrior", so the feather fan is omitted. Furthermore, some words actually have certain similar words. By segmenting the original initial description information and expanding the synonyms of the initial description information itself, it is beneficial to enrich the annotated content corresponding to the information and also to improve the search hit rate.

[0131] Furthermore, it should be noted that in the embodiments of the present application, the type of information corresponding to each tag can be different. For example, the information described above can be a video, a picture, or text. Then, if the reference tag set includes tags t1, t2, and t3, then the information corresponding to tag t1 can be a video, the information corresponding to tag t2 can be a picture, and the information corresponding to tag t3 can be text.

[0132] See also Figure 14 , Figure 14 FIG. 5 shows an information search device 500 proposed in an embodiment of the present application. The device 500 includes:

[0133] The similarity calculation unit 510 is used to obtain a reference tag set and similarity parameters corresponding to the reference tag set from multiple tags based on the similarity between the search text and the multiple tags. The similarity parameters include the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameters are sorted in order of size.

[0134] The information acquisition unit 520 is configured to acquire a reference information set corresponding to the reference tag set according to a correspondence between tags and information.

[0135] The position feature acquisition unit 530 is used to determine the position feature parameters of each information based on the sorting position of the label corresponding to each information in the reference information set in the reference label set, and the sorting position of the label in the reference label set corresponds to the sorting position of the corresponding similarity in the similarity matrix.

[0136] The information screening unit 540 is configured to obtain the information corresponding to the search text from the reference information set based on the similarity parameter and the position characteristic parameter of each information.

[0137] As a method, the information screening unit 540 is specifically used to obtain the weight parameters corresponding to the reference tag set, and the weight parameters include the weight corresponding to each tag in the reference tag set, wherein the greater the corresponding similarity of the tag, the greater the weight corresponding to the tag; based on the similarity parameter, the weight parameter and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0138] Optionally, the similarity parameter is a similarity matrix, the position feature parameter is a position feature matrix, and the weight parameter is a weight matrix. The information screening unit 540 is specifically configured to obtain the product of the similarity matrix, the weight matrix, and the position feature matrix of each piece of information as a matching value corresponding to each piece of information; and to select the information whose matching value satisfies the specified sorting condition as the information corresponding to the search text.

[0139] Optionally, the information screening unit 540 is specifically used to determine the weight corresponding to each tag based on the similarity corresponding to the tags in the reference tag set to obtain a weight parameter; wherein, the smaller the difference in the similarities corresponding to adjacent tags in the reference tag set, the greater the difference in the weights corresponding to the adjacent tags, and the adjacent tags are the two tags with the closest corresponding similarities.

[0140] As one approach, similarity calculation unit 510 is further configured to obtain a vector corresponding to the search text; calculate vector similarities between the vector corresponding to the search text and vectors of multiple tags, and use the vector similarities as the similarities between the search text and the multiple tags. Optionally, similarity calculation unit 510 is further configured to obtain a one-hot encoding corresponding to the search text; input the one-hot encoding into a target neural network, the target neural network including a hidden layer; and use weight parameters output by the hidden layer as the vector corresponding to the search text.

[0141] As a way, Figure 15 As shown, the apparatus 500 further includes: a text processing unit 550 for obtaining an input text; performing word segmentation on the input text to obtain multiple search texts. In this manner, the information screening unit 540 is further specifically configured to output the search information corresponding to each of the multiple search texts as the search information corresponding to the input text.

[0142] Optionally, the text processing unit 550 is specifically used to obtain the segmentation length, which is used to limit the upper limit of the length of the search text obtained by segmentation; the input text is segmented based on the segmentation length to obtain multiple search texts.

[0143] Optionally, the text processing unit 550 is specifically used to perform multiple word segmentation on the input text, and obtain the results of the multiple word segmentations as multiple search texts, wherein the word segmentation length corresponding to each word segmentation is different, and the word segmentation length is used to limit the upper limit of the length of the search text obtained by segmentation.

[0144] As a way, Figure 16 As shown, apparatus 500 further includes a tag processing unit 560 configured to obtain initial description information corresponding to the information; perform text processing on the initial description information to obtain processed information; and use the initial description information and the processed information as tags corresponding to the information. Optionally, the text processing includes at least one of the following processing methods: performing word segmentation on the initial description information; and obtaining synonyms corresponding to the initial description information.

[0145] The present application provides an information search device, so that through the above method, the information search device can first obtain a reference information set through the similarity between the label corresponding to the information and the search text, and then after obtaining the position feature parameters corresponding to each information and the similarity ranking, based on the position feature parameters of each information and the similarity parameters corresponding to the reference information set, filter out the information corresponding to the search text from the reference information set, thereby making the found information more relevant to the search text, so as to improve the accuracy of information search.

[0146] It should be noted that the device embodiment in this application corresponds to the aforementioned method embodiment. The specific principles in the device embodiment can be found in the contents of the aforementioned method embodiment and will not be repeated here.

[0147] The following will be combined Figure 17 An electronic device provided by this application is described.

[0148] See also Figure 17 Based on the above-mentioned information search method, the embodiments of the present application also provide another electronic device 200 including a processor 104 capable of executing the above-mentioned information search method. The electronic device 200 may be a smartphone, tablet computer, computer, or portable computer. The electronic device 200 also includes a memory 104 and a network module 106. The memory 104 stores a program capable of executing the content of the above-mentioned embodiments, and the processor 102 can execute the program stored in the memory 104.

[0149] The processor 102 may include one or more cores for processing data and a message matrix unit. The processor 102 utilizes various interfaces and circuits to connect various components within the electronic device 200. It executes instructions, programs, code sets, or instruction sets stored in the memory 104, and accesses data stored in the memory 104 to perform various functions and process data within the electronic device 200. Optionally, the processor 102 may be implemented using at least one of the following hardware forms: a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). The processor 102 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing display content; and the modem handles wireless communications. It is understood that the modem may not be integrated into the processor 102 and may be implemented separately via a communication chip.

[0150] The memory 104 may include a random access memory (RAM) or a read-only memory (ROM). The memory 104 may be used to store instructions, programs, codes, code sets, or instruction sets. The memory 104 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the following various method embodiments, etc. The data storage area may also store data created by the terminal 100 during use (such as a phone book, audio and video data, chat history data), etc.

[0151] The network module 106 is used to receive and transmit electromagnetic waves, converting electromagnetic waves into electrical signals, thereby communicating with a communication network or other devices. For example, the network module 106 can transmit broadcast data or parse broadcast data transmitted by other devices. The network module 106 may include various existing circuit components for performing these functions, such as an antenna, a radio frequency transceiver, a digital signal processor, an encryption / decryption chip, a subscriber identity module (SIM) card, a memory, and the like. The network module 106 can communicate with various networks such as the Internet, an intranet, or a wireless network, or communicate with other devices via a wireless network. The wireless network may include a cellular telephone network, a wireless local area network, or a metropolitan area network. For example, the network module 106 can exchange information with a base station.

[0152] Please refer to Figure 18 , which shows a block diagram of a computer-readable storage medium provided in an embodiment of the present application. The computer-readable medium 1100 stores program code, which can be called by a processor to execute the method described in the above method embodiment.

[0153] Computer-readable storage medium 1100 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Alternatively, computer-readable storage medium 1100 includes a non-transitory computer-readable storage medium. Computer-readable storage medium 1100 has storage space for program code 810 for executing any of the method steps described above. These program codes can be read from or written to one or more computer program products. Program code 1110 can be compressed, for example, in a suitable form.

[0154] The present application also provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the above-described information search method.

[0155] In summary, the present application provides an information search method, device, electronic device, and storage medium. After obtaining the search text, the method can obtain a reference tag set from the multiple tags based on the similarity between the search text and multiple tags, as well as similarity parameters corresponding to the reference tag set. The similarity parameters include the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameters are sorted in order of size. Then, based on the correspondence between the tags and the information, the reference information set corresponding to the reference tag set is obtained, and based on the sorting position of the tag corresponding to each information in the reference information set in the reference tag set, the position feature parameter of each information is determined. The sorting position of the tag in the reference tag set corresponds to the sorting position of the corresponding similarity in the similarity matrix. Then, based on the similarity parameters and the position feature parameter of each information, the information corresponding to the search text is obtained from the reference information set.

[0156] Therefore, through the above method, a reference information set can be obtained first through the similarity between the label corresponding to the information and the search text, and then after obtaining the position feature parameters corresponding to each information and the similarity ranking, the information corresponding to the search text can be filtered out from the reference information set based on the position feature parameters of each information and the similarity parameters corresponding to the reference information set, so that the found information is more relevant to the search text, so as to improve the accuracy of information search.

[0157] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application 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 cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An information search method, characterized in that: The method comprises: Based on the similarity between the search text and a plurality of tags, a reference tag set and a similarity parameter corresponding to the reference tag set are obtained from the plurality of tags, wherein the similarity parameter includes the similarity between each tag in the reference tag set and the search text, and the similarities in the similarity parameter are sorted in order of magnitude to obtain a similarity matrix; According to the correspondence between the label and the information, the information corresponding to each label in the reference label set is obtained, and the information corresponding to each label in the reference label set is combined into a reference information set; based on the ranking position of the label corresponding to each information in the reference information set in the reference label set, the position characteristic parameter of each information is determined, the ranking position of the label in the reference label set corresponds to the ranking position of the corresponding similarity in the similarity matrix; the position characteristic parameter corresponding to the information represents the degree of similarity between the information and the search text; Based on the weight parameters corresponding to the reference tag set, the similarity parameters and the position feature parameters of each information, the information corresponding to the search text is obtained from the reference information set; wherein, the weight parameters include the weight corresponding to each tag in the reference tag set, the greater the corresponding similarity, the greater the weight corresponding to the tag, the smaller the difference in similarities corresponding to adjacent tags in the reference tag set, the greater the difference in weights corresponding to the adjacent tags, and the adjacent tags are the two tags with the closest corresponding similarities.

2. The method according to claim 1, characterized in that The similarity parameter is a similarity matrix, the position feature parameter is a position feature matrix, and the weight parameter is a weight matrix; obtaining the information corresponding to the search text from the reference information set based on the weight parameters corresponding to the reference tag set, the similarity parameter, and the position feature parameters of each information includes: Obtaining the product of the similarity matrix, the weight matrix, and the position feature matrix of each information as a matching value corresponding to each information; The information whose corresponding matching value satisfies the specified sorting condition is used as the information corresponding to the search text.

3. The method according to claim 1, characterized in that Before obtaining a reference tag set from the multiple tags based on the similarity between the search text and the multiple tags, the method further includes: Get input text; Performing word segmentation on the input text to obtain multiple search texts; After obtaining the information corresponding to the search text from the reference information set based on the similarity parameter and the position feature parameter of each information, the method further includes: The search information corresponding to each of the plurality of search texts is output as the search information corresponding to the input text.

4. The method according to claim 3, characterized in that The input text is segmented to obtain multiple search texts, including: The input text is segmented multiple times, and the results of the multiple segmentations are obtained as multiple search texts, wherein the segmentation length corresponding to each segmentation is different, and the segmentation length is used to limit the upper limit of the length of the search text obtained by segmentation.

5. The method according to claim 1, wherein Before obtaining a reference tag set from the multiple tags based on the similarity between the search text and the multiple tags, the method further includes: Get the initial description information corresponding to the information; Performing text processing on the initial description information to obtain processed information; The initial description information and the processed information are used as labels corresponding to the information.

6. An information search device, characterized in that: The device comprises: a similarity calculation unit, configured to obtain a reference tag set and similarity parameters corresponding to the reference tag set from the multiple tags based on similarities between the search text and the multiple tags, wherein the similarity parameters include similarities between each tag in the reference tag set and the search text, and the similarities in the similarity parameters are sorted in order of magnitude to obtain a similarity matrix; an information acquisition unit, configured to acquire information corresponding to each tag in the reference tag set according to a correspondence between the tag and the information, and combine the information corresponding to each tag in the reference tag set into a reference information set; a position feature acquisition unit, configured to determine a position feature parameter of each information in the reference information set based on the ranking position of the label corresponding to each information in the reference label set, wherein the ranking position of the label in the reference label set corresponds to the ranking position of the corresponding similarity in the similarity matrix; the position feature parameter corresponding to the information represents the degree of similarity between the information and the search text; An information screening unit is used to obtain the information corresponding to the search text from the reference information set based on the weight parameters corresponding to the reference tag set, the similarity parameters and the position feature parameters of each information; wherein, the weight parameters include the weight corresponding to each tag in the reference tag set, the greater the corresponding similarity, the greater the weight corresponding to the tag, the smaller the difference in similarity corresponding to adjacent tags in the reference tag set, the greater the difference in weight corresponding to the adjacent tags, and the adjacent tags are the two tags with the closest corresponding similarity.

7. An electronic device, characterized in that: The method comprises a processor and a memory; one or more programs are stored in the memory and configured to be executed by the processor to implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores program code, wherein when the program code is executed by a processor, the method according to any one of claims 1 to 5 is executed.

Images