Data processing method, device, server and computer readable storage medium

Abstract

The present disclosure relates to a data processing method, device, server and computer-readable storage medium, wherein the data processing method comprises: obtaining search information of a target object, wherein the search information comprises at least one search text and at least one time mark corresponding to the search text; clustering the search information based on the time mark to obtain a first sequence, wherein the first sequence is composed of a plurality of elements, each of which comprises at least one search text; obtaining the degree of association between adjacent elements in the plurality of elements, wherein the plurality of elements are sorted in a preset time sequence; connecting adjacent elements whose degrees of association meet preset conditions to obtain a target element; updating the first sequence based on the target element to obtain a search sequence, wherein the search sequence is used to indicate the search behavior of the target object. The present disclosure at least solves the problem of low accuracy of the search information obtained in the related art.

Description

Technical Field

[0001] The present disclosure relates to the field of Internet technology, and in particular to a data processing method, device, server, and computer-readable storage medium. Background Art

[0002] With the development of Internet technology, search engines are used more and more widely in people's daily lives. Every day, a large number of people search for information through search engines. Among them, from the user entering the search engine to start searching to the user exiting the search engine is a complete search query process of the user in the search engine. This process can be used to train word vector models and construct collaborative filtering training samples.

[0003] However, during a complete search query, the time intervals between searches often vary greatly, and the time and text input by users are also different each time they search. This results in the inability to form effective and accurate search information during the search process, making it difficult to effectively train word vector models and collaborative filtering models, further resulting in the problem of low accuracy in obtaining search results. Summary of the invention

[0004] The present disclosure provides a data processing method, device, server and computer-readable storage medium to at least solve the problem of low accuracy of search information obtained in the related art. The technical solution of the present disclosure is as follows:

[0005] According to a first aspect of an embodiment of the present disclosure, a data processing method is provided, comprising: obtaining search information of a target object, wherein the search information includes at least one search text and at least one time identifier corresponding to the search text; clustering the search information based on the time identifier to obtain a first sequence, wherein the first sequence is composed of a plurality of elements, and each element includes at least one search text; obtaining a correlation between adjacent elements of the plurality of elements, wherein the plurality of elements are sorted according to a preset time sequence; connecting adjacent elements whose correlations meet a preset condition to obtain a target element; and updating the first sequence based on the target element to obtain a search sequence, wherein the search sequence is used to indicate the search behavior of the target object.

[0006] Furthermore, the data processing method also includes: obtaining first search information of multiple objects within a historical preset time period; clustering the first search information according to the object identifiers of the multiple objects to obtain a clustering result; obtaining a target object identifier of the target object; and determining the search information corresponding to the target object identifier in the clustering result as the search information of the target object.

[0007] Furthermore, the data processing method also includes: sorting at least one search text according to a time mark corresponding to the at least one search text to obtain a sorting result; and hierarchically clustering the sorting result according to time sequence to obtain a first sequence.

[0008] Furthermore, the data processing method also includes: obtaining a first search text from at least one search text contained in the first element in chronological order, wherein the first search text is the last text in the first element; obtaining a second search text from at least one search text contained in the second element in chronological order, wherein the first element and the second element are two temporally adjacent elements in the first sequence, the second search text is the first text in the second element, and the time stamp corresponding to the first search text is less than the time stamp corresponding to the second search text; calculating the similarity between the first search text and the second search text; and determining the correlation between the first element and the second element based on the similarity.

[0009] Furthermore, the data processing method also includes: obtaining multiple first byte fragments intercepted by the first search text sliding under the preset window and multiple second byte fragments intercepted by the second search text sliding under the preset window; calculating the first probability corresponding to the multiple first byte fragments and the second probability corresponding to the multiple second byte fragments, wherein the first probability represents the frequency of the multiple first byte fragments appearing in the first search text, and the second probability represents the frequency of the multiple second byte fragments appearing in the second search text; determining the similarity between the first search text and the second search text based on the first probability and the second probability.

[0010] Furthermore, the data processing method also includes: calculating the product of first probabilities of multiple first byte fragments to obtain a third probability corresponding to the first search text; calculating the product of second probabilities of multiple second byte fragments to obtain a fourth probability corresponding to the second search text; and obtaining the similarity between the first search text and the second search text based on the third probability and the fourth probability.

[0011] Furthermore, the data processing method also includes: when the similarity is greater than or equal to a preset similarity, obtaining the correlation between the first element and the second element; when the correlation between the first element and the second element is greater than a preset correlation, obtaining the third byte segment in the first element and the fourth byte segment in the second element, wherein the third byte segment and the fourth byte segment are byte segments that are adjacent in time; and connecting the first element and the second element according to the third byte segment and the fourth byte segment to obtain the target element.

[0012] Furthermore, the data processing method also includes: determining an element combination corresponding to the target element from multiple elements contained in the first sequence, wherein the element combination consists of a first element and a second element; and replacing the element combination in the first sequence based on the target element to obtain a search sequence.

[0013] Furthermore, the data processing method also includes: after updating the first sequence based on the target element to obtain the search sequence, training the collaborative filtering model based on the search sequence to obtain the target model, wherein the collaborative filtering model is used to push information associated with the target object based on the historical search information of the target object.

[0014] Furthermore, the data processing method also includes: after updating the first sequence based on the target element to obtain the search sequence, training the word vector model based on the search sequence to obtain the target word vector model, wherein the word vector model is used to classify the search information of the target object and push information associated with the target object based on the classification results.

[0015] According to a second aspect of an embodiment of the present disclosure, a data processing device is provided, comprising: a first acquisition unit, configured to execute acquisition of search information of a target object, wherein the search information includes at least one search text and at least one time identifier corresponding to the search text; a clustering unit, configured to execute clustering of the search information based on the time identifier to obtain a first sequence, wherein the first sequence is composed of a plurality of elements, each element including at least one search text; a second acquisition unit, configured to execute acquisition of a correlation between adjacent elements of the plurality of elements, wherein the plurality of elements are sorted in a preset time sequence; a connection unit, configured to execute connection of adjacent elements whose correlations meet a preset condition to obtain a target element; an update unit, configured to execute updating of the first sequence based on the target element to obtain a search sequence, wherein the search sequence is used to indicate the search behavior of the target object.

[0016] Further, the first acquisition unit includes: a third acquisition unit, configured to execute acquisition of first search information of multiple objects within a historical preset time length; a first clustering unit, configured to execute clustering processing of the first search information according to object identifiers of multiple objects to obtain clustering results; a fourth acquisition unit, configured to execute acquisition of the target object identifier of the target object; and a first determination unit, configured to execute determination of the search information corresponding to the target object identifier in the clustering result as the search information of the target object.

[0017] Furthermore, the clustering unit includes: a sorting unit, configured to sort at least one search text according to a time stamp corresponding to at least one search text to obtain a sorting result; and a second clustering unit, configured to perform hierarchical clustering of the sorting results in chronological order to obtain a first sequence.

[0018] Further, the second acquisition unit includes: a fifth acquisition unit, configured to execute the acquisition of a first search text from at least one search text contained in the first element in chronological order, wherein the first search text is the last text in the first element; a sixth acquisition unit, configured to execute the acquisition of a second search text from at least one search text contained in the second element in chronological order, wherein the first element and the second element are two temporally adjacent elements in the first sequence, the second search text is the first text in the second element, and the time stamp corresponding to the first search text is less than the time stamp corresponding to the second search text; a first calculation unit, configured to execute the calculation of the similarity between the first search text and the second search text; and a second determination unit, configured to execute the determination of the association between the first element and the second element based on the similarity.

[0019] Further, the first calculation unit includes: a seventh acquisition unit, configured to execute acquisition of multiple first byte fragments intercepted by the first search text sliding under the preset window and multiple second byte fragments intercepted by the second search text sliding under the preset window; a second calculation unit, configured to execute calculation of first probabilities corresponding to the multiple first byte fragments and second probabilities corresponding to the multiple second byte fragments, wherein the first probability represents the frequency of occurrence of the multiple first byte fragments in the first search text, and the second probability represents the frequency of occurrence of the multiple second byte fragments in the second search text; a third determination unit, configured to execute determination of the similarity between the first search text and the second search text based on the first probability and the second probability.

[0020] Further, the third determination unit includes: a third calculation unit, configured to calculate the product of first probabilities of multiple first byte fragments to obtain a third probability corresponding to the first search text; a fourth calculation unit, configured to calculate the product of second probabilities of multiple second byte fragments to obtain a fourth probability corresponding to the second search text; and a fifth calculation unit, configured to obtain the similarity between the first search text and the second search text based on the third probability and the fourth probability.

[0021] Further, the connection unit includes: an eighth acquisition unit, configured to execute, when the similarity is greater than or equal to a preset similarity, to obtain the correlation between the first element and the second element; a ninth acquisition unit, configured to execute, when the correlation between the first element and the second element is greater than a preset correlation, to obtain a third byte segment in the first element and a fourth byte segment in the second element, wherein the third byte segment and the fourth byte segment are byte segments that are adjacent in time; and a first connection unit, configured to execute, when the correlation between the first element and the second element is greater than a preset correlation, to connect the first element and the second element according to the third byte segment and the fourth byte segment to obtain a target element.

[0022] Furthermore, the updating unit includes: a fourth determining unit, configured to determine an element combination corresponding to a target element from a plurality of elements contained in the first sequence, wherein the element combination consists of a first element and a second element; and a replacing unit, configured to replace the element combination in the first sequence based on the target element to obtain a search sequence.

[0023] Furthermore, the data processing device also includes: a first training unit, which is configured to execute, after updating the first sequence based on the target element to obtain the search sequence, training the collaborative filtering model based on the search sequence to obtain the target model, wherein the collaborative filtering model is used to push information associated with the target object based on the historical search information of the target object.

[0024] Furthermore, the data processing device also includes: a second training unit, which is configured to execute, after updating the first sequence based on the target element to obtain the search sequence, training the word vector model based on the search sequence to obtain a target word vector model, wherein the word vector model is used to classify the search information of the target object and push information associated with the target object based on the classification result.

[0025] According to a third aspect of an embodiment of the present disclosure, a server is provided, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the above-mentioned data processing method.

[0026] According to a fourth aspect of an embodiment of the present disclosure, a computer readable storage medium is provided. When instructions in the computer readable storage medium are executed by a processor of a server, the server is enabled to execute the above-mentioned data processing method.

[0027] According to a fifth aspect of an embodiment of the present disclosure, there is provided a computer program product, comprising a computer program / instruction, wherein the computer program / instruction implements the above-mentioned data processing method when executed by a processor.

[0028] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects:

[0029] In an embodiment of the present disclosure, hierarchical clustering is performed on the search information of the target object, and the search behavior corresponding to the search information of the target object is determined based on the clustering result. After the search information of the target object is obtained, the search information is clustered based on the time mark of the search text contained in the search information to obtain a first sequence composed of multiple elements, and then the correlation between adjacent elements in the multiple elements is determined, and the adjacent elements whose correlation meets the preset conditions are connected to obtain the target element. Finally, the first sequence is updated based on the target element to obtain a search sequence representing the search behavior of the target object, wherein each element includes at least one search text, and the multiple elements are sorted according to a preset time order. In the above process, in the present disclosure, after the search information of the target object is clustered based on the time mark, the correlation between the multiple elements contained in the first sequence obtained by clustering is calculated, and then the adjacent elements in the first sequence are connected according to the correlation, thereby avoiding the problem of misclassification caused by clustering the search information based on the time mark, thereby improving the accuracy of the search information. It can be seen that the solution provided by the present disclosure achieves the purpose of accurately obtaining search information, solves the problem in the related art that the accuracy of the search information obtained is low due to different search time intervals during the search process, and achieves the effect of improving the accuracy of the search information.

[0030] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0032] Figure 1 The invention is a flowchart of a data processing method according to an exemplary embodiment.

[0033] Figure 2 is a schematic diagram of a clustering tree according to an exemplary embodiment.

[0034] Figure 3 is a schematic diagram of a first sequence according to an exemplary embodiment.

[0035] Figure 4 is a schematic diagram of a first sequence according to an exemplary embodiment.

[0036] Figure 5 is a schematic diagram of a target element according to an exemplary embodiment.

[0037] Figure 6 is a schematic diagram of a search sequence according to an exemplary embodiment.

[0038] Figure 7 It is a block diagram of a data processing device according to an exemplary embodiment.

[0039] Figure 8 It is a block diagram of a server according to an exemplary embodiment. DETAILED DESCRIPTION

[0040] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0041] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0042] It should be noted that the search engine backend can serve as the executor of the data processing method of the embodiment of the present disclosure, wherein the search engine backend can be but is not limited to: physical servers, cloud servers, desktop computers, laptops, smart tablets, smart phones and other devices.

[0043] In addition, it should be noted that the acquisition, storage and application of user personal information involved in the technical solution of the present disclosure are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0044] Figure 1 is a flow chart of a data processing method according to an exemplary embodiment. Figure 1 As shown, the data processing method is used in the search engine background and includes the following steps.

[0045] Step S11, obtaining search information of the target object.

[0046] In step S11, the search information includes at least one search text and at least one time stamp corresponding to the search text, and the target object may be a user who searches for information on the search engine. In addition, the search text may be a word that the user searches for on the search engine, such as "Chinese history", "American history", and "British history", etc.; it may also be a word or a paragraph that the user searches for. Optionally, the form of the search text includes, but is not limited to: Chinese, English, other foreign languages, and symbols. The time stamp corresponding to the search text may be the search time when the user performs a text search on the search text, and the search time may be associated with the search text in the form of a timestamp.

[0047] Step S12: clustering the search information based on the time identifier to obtain a first sequence.

[0048] In step S12, the first sequence is composed of multiple elements, each element includes at least one search text, and each search text has a corresponding time stamp. The search engine backend can cluster the search information based on the time stamp to obtain multiple elements, where the multiple elements are multiple sub-sessions, and each sub-session represents a session process in which a terminal user communicates with the interactive system.

[0049] In an optional embodiment, the search engine backend may sort the multiple search texts contained in the search information in chronological order according to the identification information corresponding to each search text, for example, the search text with an earlier time identification is before the search text with a later time identification. Further, after sorting the search texts contained in the search information based on the time identification, the search engine backend may combine a fixed number of search texts into one element, for example, 10 search texts into one element, and so on, until the remaining search texts do not meet the fixed number, then the search engine backend automatically combines the remaining search texts into one element.

[0050] In another optional embodiment, after the sorting is completed, the search engine backend may also cluster the search texts according to a fixed time length. For example, the time mark of search text 1 is 8:00, the time mark of search text 2 is 8:05, the fixed time length is set to 10 minutes, and the time difference between search text 2 and search text 1 is 5 minutes, which is less than 10 minutes. In this case, the search engine backend combines search text 1 and search text 2 into one element. If the time mark of search text 3 is 8:20, then since the time difference between search text 3 and search text 2 is greater than 10 minutes, search text 3 is not combined with search text 1 and search text 2 into one element, but search text 3 is set separately in one element.

[0051] It should be noted that when the target object performs multiple text searches, the interval between two searches may be long. In step S12, the search information is clustered based on the time identifier to obtain a first sequence consisting of multiple elements, and then the elements in the first sequence are processed to determine a search sequence that characterizes the search behavior of the target object, thereby improving the accuracy of obtaining search information.

[0052] Step S13, obtaining the correlation between adjacent elements in the plurality of elements.

[0053] In step S13, the multiple elements are sorted in the first sequence according to a preset time order, for example, the multiple elements are sorted according to the search time of the target object for the input search text. Adjacent elements are two adjacent elements of the multiple elements in the first sequence, for example, the first sequence is arranged in the order of element 1, element 2, element 3, then element 1 and element 2 are adjacent elements.

[0054] Optionally, after clustering the search information according to time stamps to obtain multiple elements, the search engine backend may further integrate the multiple elements. Since there may be correlations between the multiple elements, for example, when the user searches for the first time, the search text entered is "Beijing"; when searching for the second time, the search text entered is "Beijing", then the search texts of the user's two searches are highly correlated. However, since the search time interval when the user searches for the search information is long, the two search texts may be clustered into two different elements. In this scenario, the search engine backend can calculate the similarity of the search texts in the multiple elements, and determine the correlation between adjacent elements in the multiple elements based on the similarity.

[0055] Step S14, connecting adjacent elements whose associations meet preset conditions to obtain a target element.

[0056] In step S14, the preset condition may be, but is not limited to, a correlation greater than a preset correlation, that is, when the correlation between two adjacent elements is greater than the preset correlation, the two adjacent elements are connected to obtain the target element. For example, the first element is "tensorflow" and the second element is "tutorial". The correlation between the two elements is greater than the preset correlation, then the search engine backend connects the first element and the second element to obtain the target element "tensorflow tutorial".

[0057] It should be noted that the above-mentioned preset relevance can be set by the user according to the search experience, and can also be adjusted by the search engine backend according to the historical search information and search results.

[0058] Step S15, updating the first sequence based on the target element to obtain a search sequence.

[0059] In step S15, the search sequence is used to indicate the search behavior of the target object. Optionally, after obtaining the target element, the search engine background replaces the adjacent element corresponding to the target element in the first sequence with the target element, so as to update the first sequence and obtain the search sequence. For example, the first sequence is composed of element 1, element 2, element 3, and element 4, wherein element 1 is "tensorflow", element 2 is "tutorial", and the target element is "tensorflow tutorial". After updating the first sequence based on the target element, the search sequence is "tensorflow tutorial", element 3, and element 4.

[0060] It should be noted that in step S15, the elements in the search sequence are obtained by connecting elements with higher correlation. Since the elements with higher correlation have been connected and combined into new elements (i.e., target elements), and the new elements can more accurately represent the search behavior of the target object, the search results corresponding to the search information obtained based on the search sequence after the first sequence is updated are more accurate.

[0061] Based on the scheme defined by steps S11 to S15 of the present disclosure, it can be known that in an embodiment of the present disclosure, hierarchical clustering is performed on the search information of the target object, and the search behavior corresponding to the search information of the target object is determined based on the clustering result. After the search information of the target object is acquired, the search information is clustered based on the time mark of the search text contained in the search information to obtain a first sequence composed of multiple elements, and then the correlation between adjacent elements in the multiple elements is determined, and the adjacent elements whose correlation meets the preset conditions are connected to obtain the target element. Finally, the first sequence is updated based on the target element to obtain a search sequence representing the search behavior of the target object, wherein each element includes at least one search text, and the multiple elements are sorted in a preset time order.

[0062] It is easy to notice that in the present disclosure, after the search information of the target object is clustered based on the time identifier, the correlation between the multiple elements contained in the first sequence obtained by clustering is calculated, and then the adjacent elements in the first sequence are connected according to the correlation, thereby avoiding the problem of misclassification caused by clustering the search information based on the time identifier, thereby improving the accuracy of the search information.

[0063] It can be seen that the solution provided by the present disclosure achieves the purpose of accurately obtaining search information, solves the problem in the related art that the accuracy of the search information obtained is low due to different search time intervals during the search process, and achieves the effect of improving the accuracy of the search information.

[0064] In an optional embodiment, before clustering the search information, the search engine backend needs to obtain the search information of the target object. Specifically, the search engine backend first obtains the first search information of multiple objects within a preset historical period, and then clusters the first search information according to the object identifiers of the multiple objects to obtain a clustering result, and obtains the target object identifier of the target object, and finally determines the search information corresponding to the target object identifier in the clustering result as the search information of the target object.

[0065] Optionally, the search engine backend obtains the first search information of multiple objects within a preset historical period, for example, the search engine backend obtains the search information of multiple users within the recent month, wherein the multiple objects are users who use the search engine to search. Then, the search engine backend clusters the first search information according to the object identifier, that is, the search engine backend aggregates the search information of different objects together, for example, the search engine backend aggregates the search information of object A into cluster set 1; and aggregates the search information of object B into cluster set B, wherein the object identifier is used to distinguish different objects. Optionally, after the search engine backend clusters the first search information of multiple objects and obtains multiple cluster sets, it uses the object identifier to distinguish different cluster sets, for example, uses the object identifier of object A to identify cluster set 1, and uses the object identifier of object B to identify cluster set 2. It should be noted that the above-mentioned object identifier can be, but is not limited to, the user name, network address, user account and other information that uniquely identifies the object.

[0066] Furthermore, after clustering the first search information of multiple objects, the search engine backend can find the search information corresponding to the target object from the clustering results based on the object identifier of the target object. For example, if the object identifier of the target object is B, the search engine backend will find all the search information with the object identifier B from the clustering results, and the search information is the search information corresponding to the target object.

[0067] It should be noted that in the above process, the historical search information of each object is obtained according to the object identifier of each object, which realizes the refined acquisition of search information and improves the accuracy of obtaining search information. In addition, the search information between different objects may be different. Obtaining search information from the dimension of objects can facilitate the analysis of common information between different objects, as well as the intention analysis of search information of the same object in different time periods, which is conducive to improving the efficiency of subsequent operation management.

[0068] In an optional embodiment, after obtaining the search information of the target object, the search engine backend may cluster the search information based on the time stamp to obtain a first sequence. Specifically, the search engine backend sorts at least one search text according to the time stamp corresponding to the at least one search text to obtain a sorting result, and hierarchically clusters the sorting result according to the time sequence to obtain the first sequence.

[0069] Optionally, the search information includes at least one search text and at least one time mark corresponding to the search text, wherein the time mark corresponding to the search text may be the time corresponding to when the user searches for the text, and the time may be represented in the form of a timestamp. When there are multiple search texts in the search information, the search engine backend may sort the search texts according to the timestamps corresponding to the search texts, for example, sorting the search texts according to the order of time, with the search texts with earlier timestamps being placed before the search texts with later timestamps, thereby obtaining a sorting result of multiple search texts.

[0070] Furthermore, after sorting the search texts, the search engine backend can use a hierarchical clustering algorithm to divide the search information into multiple elements in chronological order, wherein the above-mentioned chronological order can be the arrangement order of the timestamps of multiple search texts. In addition, the hierarchical clustering algorithm is a type of clustering algorithm, which can create a hierarchical nested clustering tree by calculating the similarity between data points of different categories. In the clustering tree, the original data points of different categories are the bottom layer of the tree, and the top layer of the tree is the root node of a cluster. Through the merging algorithm of the hierarchical clustering algorithm, the two most similar search texts among all search text points can be combined by calculating the similarity between multiple search texts, and this process is iterated repeatedly until all search texts are combined, at which point the search information is divided into multiple elements. For example, in Figure 2 In the cluster tree shown, the bottom layer is the original search information A, B, C, D, E, F, G, and the 7 search text points are combined into the second layer of the cluster tree according to the similarity between the 7 search text points, namely (A, F), (B, C), (D, E) and G. By analogy, a complete hierarchical clustering tree diagram can be generated.

[0071] In addition, hierarchical clustering algorithms include but are not limited to: single linkage, complete linkage and average linkage. Among them, single linkage is a simple clustering algorithm, which can use the distance between the two closest data points in two combined data points as the distance between the two combined data points; complete linkage is a complete clustering algorithm, which can use the distance between the two farthest data points in two combined data points as the distance between the two combined data points; average linkage is an average clustering algorithm, which can calculate the distance between each data point in two combined data points and all other data points.

[0072] It should be noted that the search engine backend can select different hierarchical clustering algorithms to cluster the search text according to actual needs, which will not be elaborated in detail here.

[0073] In addition, it should be noted that, since the time interval between two text searches by the user is relatively long, when the hierarchical clustering algorithm is used to combine the search texts, the texts with a relatively long time interval may be combined, resulting in inaccurate search text acquisition. In the present disclosure, the search texts are clustered according to the timestamps, and the search information is divided into multiple elements, which avoids the influence of the search time interval on the search results and improves the accuracy of obtaining the search information.

[0074] In an optional embodiment, after clustering the search information to obtain a first sequence consisting of multiple elements, the search engine background calculates the correlation between adjacent elements in the multiple elements. Specifically, the search engine background first obtains the first search text from at least one search text contained in the first element in chronological order, and obtains the second search text from at least one search text contained in the second element in chronological order, and then calculates the similarity between the first search text and the second search text, and determines the correlation between the first element and the second element based on the similarity. Among them, the first search text is the last text in the first element; the first element and the second element are two elements that are adjacent in time in the first sequence, the second search text is the first text in the second element, and the time stamp corresponding to the first search text is less than the time stamp corresponding to the second search text.

[0075] Optionally, the above time sequence can also be the arrangement order of the timestamps of multiple elements. Since multiple elements are obtained by combining the timestamps of the search text, the timestamp corresponding to the element can be determined based on the timestamp of the search text contained in each element. For example, the average time of the timestamp of the search text contained in each element can be used as the timestamp corresponding to the current element, or the minimum or maximum value of the timestamp of the search text contained in each element can be used as the timestamp corresponding to the current element. After determining the timestamp corresponding to each element, the search engine backend sorts the multiple elements corresponding to the search information according to the timestamp corresponding to each element. For example, Figure 3 In the schematic diagram of the first sequence shown, the first sequence consists of a first element, a second element, a third element and a fourth element, and the four elements are sorted in chronological order according to their corresponding timestamps. For example, Figure 3 , the time corresponding to the first element is earlier than the time corresponding to the second element, the time corresponding to the second element is earlier than the time corresponding to the third element, and the time corresponding to the third element is earlier than the time corresponding to the fourth element.

[0076] Furthermore, since an element may contain multiple search texts, the search engine backend may select the last text in the first element, i.e., the first search text, and the first text in the second element, i.e., the second search text, respectively, in the process of obtaining the correlation between multiple elements. Among them, sorting by the timestamp of the elements, the first element is before the second element, i.e., the time stamp corresponding to the first search text is smaller than the time stamp corresponding to the second search text.

[0077] Furthermore, the search engine backend may calculate the similarity between the first search text and the second search text, and then characterize the correlation between the first element and the second element according to the similarity between the first search text and the second search text.

[0078] It should be noted that, in the above process, since the number of search texts contained in the elements may be large, performing similarity comparison on each search text in each element will greatly increase the burden on the search engine background and waste computing time. By performing similarity calculation on the first search text and the second search text, and then determining the correlation between the first element and the second element based on the similarity, the efficiency of correlation matching between multiple elements can be effectively improved and computing resources can be saved.

[0079] In an optional embodiment, after obtaining the first search text and the second search text, the search engine background can calculate the similarity between the first search text and the second search text. Specifically, the search engine background first obtains multiple first byte fragments intercepted by the first search text sliding under the preset window and multiple second byte fragments intercepted by the second search text sliding under the preset window, and calculates the first probability corresponding to the multiple first byte fragments and the second probability corresponding to the multiple second byte fragments, and then determines the similarity between the first search text and the second search text based on the first probability and the second probability. Among them, the first probability represents the frequency of the multiple first byte fragments appearing in the first search text, and the second probability represents the frequency of the multiple second byte fragments appearing in the second search text.

[0080] For example, a user enters a first search text "I want to eat Chinese food lunch spend" in a search engine at 8:02, and the number of bytes in the preset window is 2 bytes (assuming that each word is one byte). Then, the first byte segments corresponding to the first search text are "I want", "want to", "to eat", "eat Chinese", "Chinese food", "food lunch", and "lunch spend". The number of occurrences of each byte segment in the preset corpus is shown in Table 1:

[0081] Table 1

[0082] I want to eat English food lunch spend I 5 827 0 9 0 0 0 2 want 2 0 608 1 6 6 5 1 to 2 0 4 686 2 0 6 211 eat 0 0 2 0 16 2 42 0 English 1 0 0 0 0 82 1 0 food 15 0 15 0 1 4 0 0 lunch 2 0 0 0 0 1 0 0 spend 1 0 1 0 0 0 0 0

[0083] In Table 1, the first row and the second column indicate that the previous word is "I" and the current word is "want", which occurs 827 times in total.

[0084] The occurrence probability (ie, first probability) corresponding to each byte segment may be shown in Table 2:

[0085] Table 2

[0086] I want to eat English food lunch spend I 0.002 0.33 0 0.0036 0 0 0 0.00079 want 0.0022 0 0.66 0.0011 0.0065 0.0065 0.0054 0.0011 to 0.00083 0 0.0017 0.28 0.00083 0 0.0025 0.087 eat 0 0 0.0027 0 0.021 0.0027 0.056 0 English 0.0063 0 0 0 0 0.52 0.0063 0 food 0.014 0 0.014 0 0.00092 0.0037 0 0 lunch 0.0059 0 0 0 0 0.0029 0 0 spend 0.0036 0 0.0036 0 0 0 0 0

[0087] In Table 2, the first row and the second column indicate that the probability corresponding to the byte segment where the previous word is "I" and the current word is "want" is 0.33.

[0088] Similarly, the second search text is processed in the same manner as the first search text, and the second probability is calculated in the same manner as the first probability, which will not be further described herein.

[0089] It should be noted that in the process of hierarchical clustering of search information, clustering is performed based on the timestamp corresponding to the search information. When the interval between two searches by the user is long, there is a problem of mis-segmentation in hierarchical clustering of search information. Wrong segmentation of search information will reduce the accuracy of the search sequence, and thus fail to accurately push information to the user. In this application, after the search information is classified and clustered, the similarity between the two adjacent search texts is calculated based on the frequency of occurrence of byte fragments of the two adjacent search texts in the corresponding search texts, and then the correlation between adjacent elements is determined based on the similarity, and then the sequence corresponding to the search information is updated based on the correlation, thereby obtaining an accurate search sequence and ensuring the accuracy of information pushed to the user.

[0090] Optionally, the search engine background calculates the product of first probabilities of multiple first byte fragments to obtain a third probability corresponding to the first search text, calculates the product of second probabilities of multiple second byte fragments to obtain a fourth probability corresponding to the second search text, and finally, based on the third probability and the fourth probability, obtains the similarity between the first search text and the second search text.

[0091] Optionally, when calculating the similarity between the first search text and the second search text, the search engine backend may use N-gram to perform a sliding window operation of size N on the search text in the element according to the byte size, thereby forming a byte segment sequence of length N, wherein each byte segment may be called a gram, and the preset window is the above-mentioned sliding window of size N. N-gram is an algorithm based on a statistical language model.

[0092] Optionally, the search engine backend can calculate the probability of the byte segments of the first search text and the second search text through the N-gram algorithm. For example, the search engine backend counts the following byte segments and the number of occurrences: I (2533 times), want (927 times), To (2417 times), eat (746 times), Chinese (158 times), food (1093 times), lunch (341 times) and spend (278 times). In addition, the search engine backend can calculate the first probability corresponding to the above-mentioned first byte segment through the N-gram algorithm, as shown in Table 2, the frequency distribution of the byte segments.

[0093] From Table 1 and Table 2 above, we know that the byte segment "i" appears 2533 times in total, and the number of times eat appears after it is 9 times in total. The probability of this byte segment is,

[0094] p(eat|i)=p(eat,i) / p(i)=count(i,eat) / count(i)=9 / 2533=0.0036

[0095] It should be noted that the probability of a byte segment is not fixed, and may be affected by factors such as the position of the byte segment in the search text. Therefore, the first probability of the first byte segment and the second probability of the second byte segment may also be different.

[0096] Furthermore, when the first search text is s1 "i want english food" and the second search text is s2 "want i english food", the search engine backend may multiply the first probabilities of the multiple first byte segments to obtain a third probability, such as:

[0097] P(s1)=P(i| <s> )P(want|i)P(english|want)P(food|english)P(< / s> |food)

[0098] =0.25×0.33×0.0011×0.5×0.68=0.000031

[0099] And the second probabilities of the plurality of second byte segments are multiplied to obtain a fourth probability, such as:

[0100] P(s2)=P(want| <s> )P(i|want)P(english|want)P(food|english)P(< / s> |food)

[0101] =0.25*0.0022*0.0011*0.5*0.68=0.00000002057

[0102] Furthermore, the search engine backend may compare the third probability P(s1) with the fourth probability P(s2). When the two probability values ​​are closer, the corresponding first search text is more similar to the second search text.

[0103] It should be noted that in the above process, since the probabilities are calculated for multiple byte fragments separately and the probability corresponding to the search text is calculated by multiplication, the obtained probability value is very accurate, which improves the accuracy of judging the similarity between the first search text and the second search text, thereby achieving the effect of improving the accuracy of search data acquisition.

[0104] In an optional embodiment, after obtaining the correlation between adjacent elements in a plurality of elements, the search engine backend can connect adjacent elements whose correlations meet preset conditions to obtain a target element. Specifically, when the similarity is greater than or equal to a preset similarity, the correlation between the first element and the second element is obtained, and when the correlation between the first element and the second element is greater than the preset correlation, the third byte segment in the first element and the fourth byte segment in the second element are obtained, and the first element and the second element are connected according to the third byte segment and the fourth byte segment to obtain the target element. The third byte segment and the fourth byte segment are byte segments that are adjacent in time.

[0105] Optionally, the preset similarity can be obtained according to a statistical model, and the preset similarity can be obtained by calculating the probabilities corresponding to the first search text and the second search text in all elements, sorting the obtained probabilities from small to large, and then taking the 75th percentile as the preset similarity. The 75th percentile is only an example, and in actual applications, the percentile can be adjusted accordingly according to actual conditions.

[0106] Furthermore, when the similarity between the first search text and the second search text is greater than or equal to the preset similarity, if the timestamp corresponding to the first element is less than the timestamp corresponding to the second element, the search engine backend can connect the first element and the second element according to the last byte segment (i.e., the third byte segment) in the first element and the first byte segment (i.e., the fourth byte segment) in the second element to obtain the target element. Figure 4 In the schematic diagram of the first sequence shown, the first element consists of four byte segments 10, 11, 12, and 13, and the second element consists of four byte segments 20, 21, 22, and 23. If the correlation between the first element and the second element is greater than the preset correlation, the first element and the second element are connected to obtain Figure 5 Schematic diagram of the target element shown. Since the timestamps corresponding to byte segment 13 (i.e., the third byte segment) and byte segment 20 (i.e., the fourth byte segment) are adjacent in time, after the first element and the second element are connected, in the target element, byte segment 13 and byte segment 20 are adjacent.

[0107] It should be noted that since the similarity between the first search text and the second search text is greater than or equal to the preset similarity, it means that the corresponding first element and the second element are highly correlated. Therefore, connecting the first element and the second element can enrich the element content and reduce the number of elements, thereby improving the resource utilization of the search engine background. Moreover, by connecting the elements with high correlation, the relevant search information can be concentrated and distinguished from the irrelevant search information, thereby achieving the effect of improving the accuracy of obtaining search information.

[0108] In an optional embodiment, after obtaining the target element, the search engine backend updates the first sequence based on the target element to obtain a search sequence. Specifically, the search engine backend determines the element combination corresponding to the target element from the multiple elements contained in the first sequence, and replaces the element combination in the first sequence based on the target element to obtain a search sequence. The element combination consists of a first element and a second element. For example, the first sequence consists of element 1, element 2, element 3, and element 4, wherein element 1 is "tensorflow", element 2 is "tutorial", and the target element is "tensorflow tutorial", then the element combination is "element 1, element 2", and after updating the first sequence based on the target element, the search sequence is "tensorflow tutorial", element 3, and element 4. For another example, Figure 6 In the schematic diagram of the search sequence shown, the target element replaces the first and second elements in the first sequence, and we get Figure 6 The search sequence shown.

[0109] It should be noted that the first element and the second element are elements with a high correlation. The element combination obtained by combining the elements with a high correlation is the target element. Since the elements with a high correlation have been connected and combined into a new element (i.e., the target element), and the new element can more accurately represent the search behavior of the target object, therefore, the search results corresponding to the search information obtained based on the search sequence after the first sequence is updated are more accurate.

[0110] In an optional embodiment, after updating the first sequence based on the target element to obtain the search sequence, the search engine backend may train a collaborative filtering model based on the search sequence to obtain a target model, or may train a word vector model based on the search sequence to obtain a target word vector model, wherein the collaborative filtering model is used to push information associated with the target object based on the historical search information of the target object, and the word vector model is used to classify the search information of the target object and push information associated with the target object based on the classification results.

[0111] Optionally, collaborative filtering refers to an information recommendation method that uses the preferences of groups with similar interests and common experiences to recommend information that users are interested in. In the present disclosure, the search engine backend can process the historical search information of the target object based on the target model, obtain the recommended information associated with the target object output by the target model, and push the recommended information to the target object.

[0112] Optionally, the above-mentioned word vector model may include but is not limited to a series of related models for generating word vectors such as Word2vec. Among them, in the present disclosure, the search engine backend uses the word vector model to classify the search information of the target object, and pushes information to the target object according to the classification result. For example, the search engine backend determines that the type of search information input by the target object is "cosmetics" based on the search information input by the target object, and the search engine backend pushes relevant information about "cosmetics" to the target object, such as information on how to buy cosmetics, comparative information on affordable cosmetics, etc.

[0113] It should be noted that since the search sequence of the embodiment of the present disclosure is obtained by connecting multiple elements according to the correlation between the multiple elements, compared with the search sequence obtained in the related art without sorting the search information, the search sequence obtained by the embodiment of the present disclosure is more accurate, and the effect of training the word vector model is better, thereby achieving the effect of improving the training efficiency of the word vector model.

[0114] From the above content, it can be seen that in the embodiment of the present disclosure, after the search information of the target object is clustered based on the time identifier, the correlation between the multiple elements contained in the first sequence obtained by clustering is calculated, and then the adjacent elements in the first sequence are connected according to the correlation, thereby avoiding the problem of misclassification caused by clustering the search information based on the time identifier, thereby improving the accuracy of the search information.

[0115] Figure 7 is a block diagram of a data processing device according to an exemplary embodiment. Figure 7 As shown, the data processing device includes: a first acquisition unit 201, a clustering unit 202, a second acquisition unit 203, a connection unit 204 and an updating unit 205.

[0116] Among them, the first acquisition unit 201 is configured to execute the acquisition of search information of the target object, wherein the search information includes at least one search text and a time identifier corresponding to the at least one search text; the clustering unit 202 is configured to execute clustering of the search information based on the time identifier to obtain a first sequence, wherein the first sequence is composed of a plurality of elements, each element including at least one search text; the second acquisition unit 203 is configured to execute the acquisition of the correlation between adjacent elements of the plurality of elements, wherein the plurality of elements are sorted according to a preset time sequence; the connection unit 204 is configured to execute the connection of adjacent elements whose correlation meets a preset condition to obtain a target element; the updating unit 205 is configured to execute the update of the first sequence based on the target element to obtain a search sequence, wherein the search sequence is used to indicate the search behavior of the target object.

[0117] It should be noted that the first acquisition unit 201, clustering unit 202, second acquisition unit 203, connection unit 204 and update unit 205 correspond to steps S11 to S15 in the above method embodiment, and the examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the contents disclosed in the above data processing method embodiment. It should be noted that the above modules can be run in a computer terminal as part of the device.

[0118] In an optional embodiment, the first acquisition unit includes: a third acquisition unit, a first clustering unit, a fourth acquisition unit, and a first determination unit. The third acquisition unit is configured to execute the acquisition of first search information of multiple objects within a preset historical time period; the first clustering unit is configured to perform clustering processing on the first search information according to the object identifiers of multiple objects to obtain a clustering result; the fourth acquisition unit is configured to execute the acquisition of the target object identifier of the target object; and the first determination unit is configured to execute the determination of the search information corresponding to the target object identifier in the clustering result as the search information of the target object.

[0119] In an optional embodiment, the clustering unit includes: a sorting unit and a second clustering unit. The sorting unit is configured to sort at least one search text according to a time stamp corresponding to at least one search text to obtain a sorting result; and the second clustering unit is configured to perform hierarchical clustering on the sorting result according to time sequence to obtain a first sequence.

[0120] In an optional embodiment, the second acquisition unit includes: a fifth acquisition unit, a sixth acquisition unit, a first calculation unit, and a second determination unit. The fifth acquisition unit is configured to acquire the first search text from at least one search text contained in the first element in chronological order, wherein the first search text is the last text in the first element; the sixth acquisition unit is configured to acquire the second search text from at least one search text contained in the second element in chronological order, wherein the first element and the second element are two temporally adjacent elements in the first sequence, the second search text is the first text in the second element, and the time mark corresponding to the first search text is less than the time mark corresponding to the second search text; the first calculation unit is configured to calculate the similarity between the first search text and the second search text; the second determination unit is configured to determine the association between the first element and the second element according to the similarity.

[0121] In an optional embodiment, the first calculation unit includes: a seventh acquisition unit, a second calculation unit and a third determination unit. The seventh acquisition unit is configured to execute acquisition of multiple first byte segments intercepted by the first search text sliding under the preset window and multiple second byte segments intercepted by the second search text sliding under the preset window; the second calculation unit is configured to execute calculation of the first probability corresponding to the multiple first byte segments and the second probability corresponding to the multiple second byte segments, wherein the first probability represents the frequency of the multiple first byte segments appearing in the first search text, and the second probability represents the frequency of the multiple second byte segments appearing in the second search text; the third determination unit is configured to execute determination of the similarity between the first search text and the second search text based on the first probability and the second probability.

[0122] In an optional embodiment, the third determination unit includes: a third calculation unit, a fourth calculation unit, and a fifth calculation unit. The third calculation unit is configured to calculate the product of the first probabilities of the plurality of first byte segments to obtain the third probability corresponding to the first search text; the fourth calculation unit is configured to calculate the product of the second probabilities of the plurality of second byte segments to obtain the fourth probability corresponding to the second search text; and the fifth calculation unit is configured to obtain the similarity between the first search text and the second search text based on the third probability and the fourth probability.

[0123] In an optional embodiment, the connection unit includes: an eighth acquisition unit, a ninth acquisition unit, and a first connection unit. The eighth acquisition unit is configured to acquire the correlation between the first element and the second element when the similarity is greater than or equal to the preset similarity; the ninth acquisition unit is configured to acquire the third byte segment in the first element and the fourth byte segment in the second element when the correlation between the first element and the second element is greater than the preset correlation, wherein the third byte segment and the fourth byte segment are byte segments adjacent in time; the first connection unit is configured to connect the first element and the second element according to the third byte segment and the fourth byte segment to obtain the target element.

[0124] In an optional embodiment, the updating unit includes: a fourth determining unit and a replacing unit. The fourth determining unit is configured to determine an element combination corresponding to a target element from a plurality of elements included in the first sequence, wherein the element combination consists of a first element and a second element; and the replacing unit is configured to replace the element combination in the first sequence based on the target element to obtain a search sequence.

[0125] In an optional embodiment, the data processing device also includes: a first training unit, configured to execute, after updating the first sequence based on the target element to obtain the search sequence, training the collaborative filtering model based on the search sequence to obtain the target model, wherein the collaborative filtering model is used to push information associated with the target object based on the historical search information of the target object.

[0126] In an optional embodiment, the data processing device also includes: a second training unit, which is configured to execute, after updating the first sequence based on the target element to obtain the search sequence, training the word vector model based on the search sequence to obtain a target word vector model, wherein the word vector model is used to classify the search information of the target object and push information associated with the target object based on the classification result.

[0127] It should be noted that, regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment of the method, and will not be elaborated here.

[0128] According to an embodiment of the present disclosure, there is also provided an embodiment of a server, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement any one of the data processing methods described above.

[0129] Optionally, Figure 8 is a block diagram of a server according to an exemplary embodiment. Figure 8 As shown, the computer terminal may include: at least one (only one is shown in the figure) processor 31, a memory 32 and a peripheral interface 33, etc.

[0130] Among them, the memory can be used to store software programs and modules, such as program instructions / modules corresponding to the data processing method and device in the embodiment of the present disclosure. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned data processing method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0131] The processor can call the information and application programs stored in the memory through the transmission device to execute any of the above-mentioned data processing method steps.

[0132] According to an embodiment of the present disclosure, an embodiment of a computer-readable storage medium is also provided. When the instructions in the above-mentioned computer-readable storage medium are executed by a processor of a server, the server is enabled to execute any one of the above-mentioned data processing methods.

[0133] According to an embodiment of the present disclosure, there is provided an embodiment of a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes any one of the above-mentioned data processing methods.

[0134] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0135] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0136] In the several embodiments provided in the present disclosure, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the above-mentioned units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0137] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0138] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0139] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable computer storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the relevant technology or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a computer storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the above-mentioned methods of each embodiment of the present invention. The aforementioned computer storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0140] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

[0141] In addition, those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0142] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A data processing method, characterized in that: include: Acquire search information of the target object, wherein the search information includes at least one search text and a time mark corresponding to the at least one search text; Clustering the search information based on the time identifier to obtain a first sequence, wherein the first sequence is composed of a plurality of elements, each element including at least one search text; Obtaining a correlation degree between adjacent elements in the plurality of elements, wherein the plurality of elements are sorted in a preset time sequence; Connecting adjacent elements whose correlations satisfy preset conditions to obtain a target element; updating the first sequence based on the target element to obtain a search sequence, wherein the search sequence is used to indicate a search behavior of the target object; Wherein, the obtaining of the correlation between adjacent elements in the multiple elements includes: obtaining a first search text from at least one search text contained in a first element according to the time sequence, wherein the first search text is the last text in the first element; obtaining a second search text from at least one search text contained in a second element according to the time sequence, wherein the first element and the second element are two temporally adjacent elements in the first sequence, the second search text is the first text in the second element, and the time identifier corresponding to the first search text is smaller than the time identifier corresponding to the second search text; calculating the similarity between the first search text and the second search text; and determining the correlation between the first element and the second element according to the similarity.

2. The data processing method according to claim 1, characterized in that: The calculating the similarity between the first search text and the second search text includes: Acquire a plurality of first byte segments captured when the first search text slides under a preset window and a plurality of second byte segments captured when the second search text slides under the preset window; Calculating first probabilities corresponding to the plurality of first byte segments and second probabilities corresponding to the plurality of second byte segments, wherein the first probability represents the frequency of occurrence of the plurality of first byte segments in the first search text, and the second probability represents the frequency of occurrence of the plurality of second byte segments in the second search text; The similarity between the first search text and the second search text is determined according to the first probability and the second probability.

3. The data processing method according to claim 2, characterized in that: The determining the similarity between the first search text and the second search text according to the first probability and the second probability includes: Calculate the product of the first probabilities of the multiple first byte segments to obtain a third probability corresponding to the first search text; Calculate the product of the second probabilities of the plurality of second byte segments to obtain a fourth probability corresponding to the second search text; The similarity between the first search text and the second search text is obtained according to the third probability and the fourth probability.

4. The data processing method according to claim 1, characterized in that: The step of connecting adjacent elements whose associations satisfy a preset condition to obtain a target element includes: When the similarity is greater than or equal to a preset similarity, obtaining a correlation between the first element and the second element; When the correlation between the first element and the second element is greater than a preset correlation, obtaining a third byte segment in the first element and a fourth byte segment in the second element, wherein the third byte segment and the fourth byte segment are byte segments adjacent in time; The first element and the second element are connected according to the third byte segment and the fourth byte segment to obtain the target element.

5. The data processing method according to claim 4, characterized in that: The updating of the first sequence based on the target element to obtain a search sequence includes: Determine an element combination corresponding to the target element from a plurality of elements included in the first sequence, wherein the element combination consists of the first element and the second element; The combination of elements in the first sequence is replaced based on the target element to obtain the search sequence.

6. The data processing method according to claim 1, characterized in that: The step of obtaining the search information of the target object includes: Obtaining the first search information of multiple objects within a preset historical time period; Performing clustering processing on the first search information according to the object identifiers of the multiple objects to obtain a clustering result; Obtain a target object identifier of the target object; The search information corresponding to the target object identifier in the clustering result is determined as the search information of the target object.

7. The data processing method according to claim 1, characterized in that: The clustering the search information based on the time identifier to obtain a first sequence includes: Sorting the at least one search text according to the time identifier corresponding to the at least one search text to obtain a sorting result; The sorting results are hierarchically clustered according to time sequence to obtain the first sequence.

8. The data processing method according to claim 1, characterized in that: After the first sequence is updated based on the target element to obtain a search sequence, the method further includes: The collaborative filtering model is trained based on the search sequence to obtain a target model, wherein the collaborative filtering model is used to push information associated with the target object according to historical search information of the target object.

9. The data processing method according to claim 1, characterized in that: After the first sequence is updated based on the target element to obtain a search sequence, the method further includes: The word vector model is trained based on the search sequence to obtain a target word vector model, wherein the word vector model is used to classify the search information of the target object and push information associated with the target object according to the classification result.

10. A data processing device, characterized in that: include: A first acquisition unit is configured to acquire search information of a target object, wherein the search information includes at least one search text and a time mark corresponding to the at least one search text; A clustering unit, configured to perform clustering on the search information based on the time identifier to obtain a first sequence, wherein the first sequence is composed of a plurality of elements, each element including at least one search text; A second acquisition unit is configured to acquire the association degree between adjacent elements in the plurality of elements, wherein the plurality of elements are sorted according to a preset time sequence; A connection unit, configured to connect adjacent elements whose correlations meet a preset condition to obtain a target element; an updating unit, configured to update the first sequence based on the target element to obtain a search sequence, wherein the search sequence is used to indicate a search behavior of the target object; Wherein, the second acquisition unit includes: a fifth acquisition unit, configured to execute acquisition of a first search text from at least one search text contained in a first element in chronological order, wherein the first search text is the last text in the first element; a sixth acquisition unit, configured to execute acquisition of a second search text from at least one search text contained in a second element in chronological order, wherein the first element and the second element are two temporally adjacent elements in the first sequence, the second search text is the first text in the second element, and the time identifier corresponding to the first search text is less than the time identifier corresponding to the second search text; a first calculation unit, configured to execute calculation of the similarity between the first search text and the second search text; and a second determination unit, configured to determine the association between the first element and the second element according to the similarity.

11. The data processing device according to claim 10, characterized in that: The first computing unit comprises: A seventh acquisition unit is configured to acquire a plurality of first byte segments captured when the first search text slides under a preset window and a plurality of second byte segments captured when the second search text slides under the preset window; A second calculation unit is configured to calculate first probabilities corresponding to the plurality of first byte segments and second probabilities corresponding to the plurality of second byte segments, wherein the first probability represents the frequency of occurrence of the plurality of first byte segments in the first search text, and the second probability represents the frequency of occurrence of the plurality of second byte segments in the second search text; A third determining unit is configured to determine the similarity between the first search text and the second search text according to the first probability and the second probability.

12. The data processing device according to claim 11, characterized in that: The third determining unit includes: A third calculation unit is configured to calculate the product of the first probabilities of the plurality of first byte segments to obtain a third probability corresponding to the first search text; a fourth calculation unit, configured to calculate the product of the second probabilities of the plurality of second byte segments to obtain a fourth probability corresponding to the second search text; The fifth calculation unit is configured to obtain the similarity between the first search text and the second search text according to the third probability and the fourth probability.

13. The data processing device according to claim 10, characterized in that: The connecting unit comprises: An eighth acquisition unit is configured to acquire the association degree between the first element and the second element when the similarity is greater than or equal to a preset similarity; a ninth acquiring unit, configured to acquire, when the correlation between the first element and the second element is greater than a preset correlation, a third byte segment in the first element and a fourth byte segment in the second element, wherein the third byte segment and the fourth byte segment are byte segments adjacent in time; The first connection unit is configured to connect the first element and the second element according to the third byte segment and the fourth byte segment to obtain the target element.

14. The data processing device according to claim 13, characterized in that: The updating unit comprises: a fourth determining unit, configured to determine an element combination corresponding to the target element from a plurality of elements included in the first sequence, wherein the element combination consists of the first element and the second element; A replacement unit is configured to replace a combination of elements in the first sequence based on the target element to obtain the search sequence.

15. The data processing device according to claim 10, characterized in that: The first acquiring unit includes: A third acquisition unit is configured to execute acquisition of first search information of multiple objects within a preset historical time period; A first clustering unit is configured to perform clustering processing on the first search information according to the object identifiers of the multiple objects to obtain a clustering result; A fourth acquisition unit is configured to execute acquisition of a target object identifier of the target object; The first determining unit is configured to determine the search information corresponding to the target object identifier in the clustering result as the search information of the target object.

16. The data processing device according to claim 10, characterized in that: The clustering unit comprises: A sorting unit, configured to sort the at least one search text according to a time identifier corresponding to the at least one search text to obtain a sorting result; The second clustering unit is configured to perform hierarchical clustering on the sorting results in chronological order to obtain the first sequence.

17. The data processing device according to claim 10, characterized in that: The data processing device further comprises: The first training unit is configured to execute, after updating the first sequence based on the target element to obtain a search sequence, training the collaborative filtering model based on the search sequence to obtain a target model, wherein the collaborative filtering model is used to push information associated with the target object based on historical search information of the target object.

18. The data processing device according to claim 10, characterized in that: The data processing device further comprises: The second training unit is configured to execute, after the first sequence is updated based on the target element to obtain a search sequence, training the word vector model based on the search sequence to obtain a target word vector model, wherein the word vector model is used to classify the search information of the target object and push information associated with the target object according to the classification result.

19. A server, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the data processing method according to any one of claims 1 to 10.

20. A computer-readable storage medium, when instructions in the computer-readable storage medium are executed by a processor of a server, enables the server to execute the data processing method according to any one of claims 1 to 9.

21. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the data processing method according to any one of claims 1 to 9 is implemented.

Images