WO2019056661A1

WO2019056661A1 - Search term pushing method and device, and terminal

Info

Publication number: WO2019056661A1
Application number: PCT/CN2018/071860
Authority: WO
Inventors: 米献艳; 王粲
Original assignee: 北京小度信息科技有限公司
Priority date: 2017-09-25
Filing date: 2018-01-09
Publication date: 2019-03-28
Also published as: CN107798066A

Abstract

The present invention relates to the field of data processing technology and discloses a search term pushing method and device, and a terminal. The method is used for pushing search terms, and specifically comprises the following steps: obtaining identity information and location information of a user and current time information (101); extracting popular search terms of the user from user historical behavior data corresponding to the identity information, so as to obtain a first search term set (102); obtaining general user trending search terms corresponding to the location information and the time information, so as to obtain a second search term set (103), wherein the general user trending search terms are search terms used in search operations performed by general users at a location and time corresponding to the location information and the time information; screening a preset number of search terms to be pushed from the first search term set and the second search term set respectively according to a search term ranking (104); and pushing the search terms to be pushed to the user (105). The present invention can push search terms specifically for different users so as to increase click-through rates on pushed trending search terms.

Description

Search word pushing method, device and terminal

Cross-reference to related applications

The present application claims the priority of the Chinese Patent Application No. JP-A-------

Technical field

The embodiments of the present disclosure relate to the field of data processing technologies, and in particular, to a search term pushing method, device, and terminal.

Background technique

With the popularity of the Internet, people are becoming more accustomed to dealing with real-life problems through the Internet, especially on people's basic life issues, such as ordering food online. By registering a member on the ordering platform, the user can realize a series of operations such as selecting a meal, ordering a meal, paying for a comment, and commenting through the ordering platform, which is very convenient.

In the ordering application provided by the current ordering platform to the user, a search bar is generally set at the head of the page for the user to directly search for the relevant shop or dish, and in the search bar or in the secondary page of the search bar, Users provide some popular search terms for users to click directly for quick search. However, the existing search word push method often only pushes the popular search words according to the popular popular search words counted by the platform or according to the sales data and presentation data of each store in the platform, and the search word push results are not targeted. The user's needs are not met, and the probability that the user clicks on the popular search term is not high. When the user queries the required content, the user needs to perform an input search to realize the query of the required content, so that the user's query efficiency is lowered.

Summary of the invention

In view of the above problems, embodiments of the present disclosure provide a search term pushing method, apparatus, and terminal, the main purpose of which is to specifically push search terms for different users, and improve the click rate of pushing popular search words.

In order to solve the above technical problem, in a first aspect, an embodiment of the present disclosure provides a search term pushing method, where the method includes:

Obtain the user's identity information, location information, and current time information;

Extracting a hot search term corresponding to the user from the user historical behavior data corresponding to the identity information, to obtain a first search term set;

Obtaining a public search term corresponding to the location information and the time information, to obtain a second search term set, wherein the public hot search term is used by a mass user to perform a search operation at a location and time corresponding to the location information and the time information. Search term;

Sorting a preset number of to-be-pushed search words from the first search word set and the second search word set respectively according to the heat ranking of the search words;

Pushing the to-be pushed search term to the user.

In a second aspect, an embodiment of the present disclosure provides a search term pushing device, where the device includes:

An obtaining module, configured to acquire identity information, location information, and current time information of the user;

a first extraction module, configured to extract, from the user historical behavior data corresponding to the identity information acquired by the acquiring module, a user hot search term corresponding to the user, to obtain a first search term set;

a second extraction module, configured to acquire a public hot search term corresponding to the location information acquired by the acquiring module, and obtain a second search term set, where the public hot search term is used by the mass user in the location information and time The search term used by the search operation for the location and time corresponding to the information;

a screening module, configured to filter a preset number of to-be-pushed from the first search word set extracted by the first extraction module and the second search word set extracted by the second extraction module, respectively, according to the heat ranking of the search words Search word

And a pushing module, configured to push the to-be-pushed search word selected by the screening module to the user.

In a third aspect, an embodiment of the present disclosure provides a search term pushing terminal, where the terminal includes a processor and a memory, wherein the memory is configured to store one or more computer instructions, and the one or more computer instructions are The processor executes steps to implement the search word push method described above.

In a fourth aspect, an embodiment of the present disclosure provides a computer readable storage medium having stored thereon computer instructions, wherein the steps of the search word pushing method described above are implemented when the computer instructions are executed by a processor.

A search word pushing method, device and terminal according to an embodiment of the present disclosure enable a user to push a targeted one based on the user's current time, location, and user identity while entering the ordering application. Batch search terms. Among these search terms, not only the hot search term based on the historical behavior of the user, but also the hot search term based on the user's current location and time statistics, so that the user is recommended. The search term covers search terms of different dimensions such as time, region and user preferences, which improves the probability that the user can find the desired content in the recommended search words, thereby achieving the purpose of improving the click rate of the popular search words. , allowing users to quickly query the content they need. At the same time, for different users, because the user's identity is different, and the application time and location are not exactly the same, therefore, different users get different push search words when they use, which realizes different users. Targeted search terms.

DRAWINGS

In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below. Obviously, the drawings in the following description It is a certain embodiment of the present disclosure, and other drawings may be obtained from those skilled in the art without any inventive effort.

FIG. 1 is a flowchart of a search word pushing method according to an embodiment of the present disclosure;

FIG. 2 is a flowchart of another search word pushing method according to an embodiment of the present disclosure;

FIG. 3A is a diagram showing a display effect of a search term in an application first page according to an embodiment of the present disclosure; FIG.

FIG. 3B is a diagram showing a display effect of a search term in a secondary page of an application according to an embodiment of the present disclosure; FIG.

FIG. 4 is a flowchart of a method for extracting search words in a first search term set according to an embodiment of the present disclosure;

FIG. 5 is a schematic flowchart diagram of a search word pushing method according to an embodiment of the present disclosure;

FIG. 6 is a structural block diagram of a search term pushing device according to an embodiment of the present disclosure;

FIG. 7 is a structural block diagram of another search term pushing apparatus according to an embodiment of the present disclosure.

Detailed ways

The technical solutions in the embodiments of the present disclosure will be clearly and completely described in conjunction with the drawings in the embodiments of the present disclosure. It is a partial embodiment of the present disclosure, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present disclosure without departing from the inventive scope are the scope of the disclosure.

The embodiment of the present disclosure provides a search word pushing method. As shown in FIG. 1 , the method should be mainly used to push a number of search words to a user by displaying a page, and the search words are calculated and more in line with the current application of the user. The vocabulary of the scene, thereby eliminating the process of the user constructing the search term, and achieving the purpose of quick query and retrieval by directly clicking the pushed search word. The specific steps of the method include:

101. Obtain identity information, location information, and current time information of the user.

In the embodiment of the present disclosure, the user uses the application program in the smart terminal as an example. The smart terminal may be a terminal device such as a mobile phone or a computer. The first step when the user is using the application is generally to open the application and enter the first page of the application. In the first page, a search function area for query operation is set, and the user can implement the operation by clicking the operation of the area. Specific search capabilities in the app. Specifically, in this step, when it is determined that a user opens and enters the application, the application or the server connected to the application will acquire identity information, location information, and current time information of the user.

The identity information is related information of the user who can use the application, and is not limited to the registered user information or the user device information. Generally, when the user enters the application, the login mode mainly includes the registration login and the visitor. Login, the difference is whether the user has entered the necessary user registration information set by the application. The registered user is a registered user, and for the un-entered, the application will automatically assign a unique identity information to the user to distinguish it, and the identity information is generally generated based on the device information of the device used by the user. The user who uses the device by default is the same user when not registered for login. For the identity information of different users, the application records the behavior data of the user separately with the behavior data of other users, so as to facilitate subsequent lookup calls.

In addition, the location information and the current time information are based on the current usage of the user, and the current location and time of the user are obtained. Generally, the existing smart terminal has a positioning function, and the specific location information is not limited to satellite positioning or network. The base station is located, and the time information can also be obtained according to the time information provided in the device. In a general application, the location and time information may have user-defined settings, and in this step, the location information and the time information are based on the current actual location and time information of the user acquired by the smart device. This can be used as a screening condition for the search term to be pushed.

102. Extract a user hot search term corresponding to the user from the user historical behavior data corresponding to the identity information, to obtain a first search term set.

Based on the user identity information obtained in step 101 above, the application extracts historical historical behavior data of the user in the application according to the identity information of the user, wherein the historical behavior data considers the problem of occupying the data storage space. The historical behavior data recorded generally has a certain time period, for example, user behavior data in the past one year or nearly three years.

In this step, the historical behavior data of the user refers to a specific operation that the user has performed in the application, and does not limit the specific operation content. However, for the recorded historical behavior data, each article has its corresponding identifier. For example, for the user click operation, the name of the clicked object is the identifier of the click operation, and for the search operation, the search entered during the search operation. The word is the identifier of the search operation. Then, when the user has corresponding user historical behavior data, the historical behavior data correspondingly has corresponding identifiers, and the identifiers are the hot words of the user searched in the first search term set.

In the process of extracting the user search term to form the first search word set, the principle of screening the identifier is based on the user's operation heat, and the specific criterion may be based on time, that is, the closer the operation is, the higher the heat corresponding to the operation. It may be based on the number of times, that is, the higher the number of times the user operates, the higher the degree of heat, and the comprehensive screening of indicators of multiple dimensions such as the fusion time and the number of times. Thereby, a plurality of identifiers are obtained as the user hot search words corresponding to the user based on the historical behavior data. The number of the user's hot search words, or the number of the user's historical behavior data, is customized according to the specific application scenario in this step, that is, in different scenarios, the first search word set may be The number of users in the hot search is set.

It can be seen that the user hot search term included in the first search word set reflects the preference and characteristics of the user performing the operation using the application, which is more suitable for the user's operation preference.

103. Obtain a public search term corresponding to the location information and the time information, and obtain a second search term set.

The public hot search term in this step refers to a search term used by a mass user to perform a search operation at a location and time corresponding to the location information and the time information. The public user here refers to all users who use the application that contains the user. It can be seen that the search term recorded in the second search term set is a search term corresponding to all user search operations integrated in the application, which reflects the search preferences of the mass user.

In addition, the location information and the time information in this step are the location information of the user acquired in step 101 and the current time information, then the public hot search term in the second search term set is based on the location information and the time information. The change in the change, the different location information may define the operation behavior of different locations or regions, that is, the mass user is a search operation performed in the location or the region, and the different time information may define the operation behavior at different times or time periods. That is, the mass user is a search operation performed at that point in time or time period. Therefore, the second set of search words in this step is a set of search words corresponding to the public user performing the search operation that defines the location and time.

Similarly, in this step, the number of popular hot search words in the second search term set may also change according to the change of the application scenario, that is, the number of popular hot search words is not specifically limited, Customize as needed.

104. Sort the preset number of to-be-pushed search words from the first search word set and the second search word set according to the heat ranking of the search words.

Since the first search word set and the second search word set respectively represent the user's own preferences and the public's preferences, the search words in the two sets can be regarded as predicting the search content that the current user may need from different dimensions. There is no correlation between the two in the search term content. To this end, this step extracts a preset number of search words from the two sets as the search words to be pushed, respectively.

It should be noted that the search words to be pushed from the two sets are different. Here, although the search words in the two sets are not related, there is still a possibility that the search words are the same in both. Therefore, it is necessary to avoid repeated extraction when extracting search words to be pushed. In addition, when extracting the search term to be pushed, it is based on the popularity ranking in each search term set. For this reason, when the first search word set and the second search word set are obtained, the search recorded in each set is obtained. The word also needs to be sorted by heat. For the way of hot sorting, this step does not limit the specific sorting method according to the number of times the user executes or the time of execution, or the way of using multiple dimensions for comprehensive sorting.

What can be obtained by this step is a plurality of hot search words based on the user's own preferences and a plurality of hot search words based on the preferences of the mass user under the limitation of specific location information and time information. The real demand of the user is likely to be one or more of these hot search words. Therefore, using these hot search words as the search words to be pushed, realizing the user's direct click search operation becomes a high probability event.

105. Push the search word to be pushed to the corresponding user.

The operations of steps 101 to 104 above are based on the background execution of the application, and after obtaining a plurality of search words to be pushed, the search terms to be pushed need to be displayed in the display interface of the application. Generally, the search area in the application includes at least a user input search word input area, and some display areas with search words. For the display interface where only the search term input area exists, the search word to be pushed may be displayed in the input area in a grayscale manner for the user to view and click the operation; and for the display interface in which the search word display area exists, The search terms to be pushed are displayed directly in the display area for the user to click.

The search word pushing method adopted by the embodiment of the present disclosure can realize that when the user enters the application, the application will push based on the current time, location and user identity of the user operation. Targeted search terms. Among these search words, not only the hot search term based on the historical behavior of the user, but also the hot search term based on the user's location and the current time statistics, so that the user is The recommended search terms cover time, geography, and user preferences, public preferences, and other different dimensions of search terms, which improves the probability that users can find the desired content in the recommended search terms, thereby improving the push of popular search terms. The purpose of the click rate is to allow users to quickly query the content they need. At the same time, for different users, because the user's identity is different, and the application time and location are not exactly the same, therefore, different users get different push search words when they use, which realizes different users. Targeted search terms.

In order to explain in more detail the search term pushing method proposed by the embodiment of the present disclosure, particularly in the ordering application, the personalized search term is pushed to the user, so that the user can quickly find the specific content to be searched. The specific steps are shown in Figure 2, including:

201. Obtain identity information, location information, and current time information of the user.

In this embodiment, the main purpose of the user to use the ordering application is to select a desired store for ordering, and in the process of ordering, in addition to the quality and price of the dish, the speed of the meal is also considered by the user. index. The speed of the meal delivery is related to the user's current location and time. For the location, the closer the store is to the user's location, the faster the meal delivery may arrive, and for time, it is directly related to the store order. The number of users, that is, the meal delivery rate during the peak meal hours, is lower than the peak period. Therefore, in the ordering application, the location information of the user and the current time information have a correlation with whether the user places an order, and when the relevant search term is pushed to the user, the main purpose is also to facilitate the effective ordering of the user, so, push The search term needs to consider the current location and time of the user.

The specific information about the identity information, the location information, and the current time information is described in step 101 in the foregoing embodiment, and details are not described herein again.

202. Determine whether the current user has user historical behavior data in the preset geographical area and the preset time period.

The preset geographical range is associated with the location information, and the preset time period is associated with the current time information. That is, determining a preset geographical area including the location information and a preset time period including the current time information according to the acquired location information and current time information, and querying the user within the preset geographical area. And whether the corresponding operation has been performed within the preset time period. Because the operation performed under this condition has a high probability that the user repeats the same operation, based on the judgment of the condition, when it is determined that the user has corresponding historical behavior data, the priority is extracted from the historical behavior data. Corresponding user hot search terms. If there is no corresponding historical behavior data for the user, the qualification condition for extracting the historical behavior data of the user needs to be extended. The specific extended rule is not specifically limited in this step, and the preset may be gradually extended. If there is no corresponding user history behavior data, the preset time period is gradually extended; of course, the reverse gradual expansion may also be performed, and the range of the preset preset area and the preset time period may be simultaneously extended to obtain the corresponding User history behavior data. If the user is a new user and does not have the user history behavior data, the historical behavior data of the user does not need to be acquired again, that is, there is no corresponding user hot search term in the first search term set extracted by the user. Conversely, if the user has historical behavior data, step 203 is performed.

203. Extract a user hot search term corresponding to the user from the user historical behavior data corresponding to the identity information, to obtain a first search term set.

For the specific implementation of this step, the manner of extracting the user's hot search term includes:

First, the historical behavior data of the user is obtained according to the identity information of the user, wherein the specific description has been made in step 202, and the manner of acquiring the historical behavior data within a specific range is specifically limited. For the obtained historical behavior data, in this embodiment, the data generated by the user performing the click behavior, the search behavior, and the access behavior may be the click behavior of the user viewing the merchant or viewing the specific food in the merchant page. The search behavior may be a search operation performed by the user in the application interface or the merchant page, and the access behavior may be an operation of the user to browse the user evaluation information of the merchant or the dish. In addition, the content of the historical behavior data may also include other activities such as shopping carts, collections, etc., in order to expand and enrich the user's favorite features.

Secondly, the object name of the user performing the operation is filtered from the historical behavior data by using a preset rule. The preset rule refers to a specific rule for extracting the object name of the user operation object, and the object name obtained by the different preset rules may be different, for example, the click behavior of the user who clicks on the B item in the A store. When the preset rule is to extract only the dish name, the obtained object name is "Dish B", and when the preset rule is to extract the shop name and the dish name, the obtained object name is "Dish B in the shop A". This step is not specifically limited for the setting of this rule.

Finally, the vocabulary whose similarity with the object name reaches the threshold is determined as the user's hot search term.

Wherein, when calculating the similarity of the object name, the specific similarity calculation method is not limited in this step, and the commonly used calculation method includes the pre-selected similarity calculation, and the other similar measurement methods may also be used to calculate the user hot search term. Such as association rules, item-based collaborative filtering, restricted Boltzmann machine recommendation algorithms, etc. For the threshold of comparison similarity, the setting needs to be adjusted according to the specific application, that is, the value of the threshold is generally an empirical value, and needs to be manually adjusted to obtain a sufficient number of user hot search words that match the similarity degree. .

It can be seen that the user hot search term in the first search term set in this embodiment includes an object name corresponding to the historical behavior data of the user, and a vocabulary similar to the object name. Therefore, even if the user has only one historical behavior data, based on the object name corresponding thereto, a plurality of user hot search words can be extended by the similarity calculation to ensure that the first search term set has multiple optional user hot searches. word.

204. Obtain a public search term corresponding to the location information and the time information, to obtain a second search term set.

The specific process of obtaining popular hot search terms in this step includes:

First, use the current location information to obtain popular hot search terms within the preset geographic area.

Second, the current time information is used to obtain the popular hot search term within the preset time period.

Thirdly, the popular hot search words in the preset geographical area and the popular hot search words in the preset time period are sorted in descending order according to the number of searches;

Fourth, extracting a preset number of popular hot search words according to the order from high to low in the descending order.

The foregoing first and second steps do not distinguish the sequence in the execution, and based on the foregoing specific execution process, the step needs to preset a preset geographical range and a preset time period, wherein the preset geographical range is an area. According to certain rules, it can be divided into several blocks. For example, in Beijing, it can be divided into Dongcheng District, Xicheng District, Chaoyang District, Haidian District, etc. according to the administrative division. According to the business circle, it can be divided into Guomao Business Circle and Zhongguancun. Business district, Asian Games Village business district, etc. The preset time period is to divide the day into a plurality of time periods according to the time of eating. For example, in this embodiment, 6 to 9 o'clock can be set as the breakfast time, 11 to 14 o'clock as the lunch time, and 14 to 16 o'clock. For afternoon tea, 17 to 20 for dinner, and 21 to 24 for 5 different hours.

When the user starts to use the ordering application, according to the location information and the time information, the corresponding public hot search term is obtained, that is, the preset geographical range where the user is currently located and the corresponding preset time period are determined, thereby filtering the pre-prepared time. In the geographical scope, and the search words used by the public users in the preset time period, the search words are sorted in descending order according to the number of search times of the application, and multiple search words are selected from the high end as the popular hot search words. It should be pointed out that in the process, the search words that meet the preset geographical range and the preset time period need to be preferentially selected as the popular hot search words, and are preferentially ranked in the subsequent sorting, for example, there is A, B, C, D, E, F six popular hot search words, wherein A and B are search terms that match the preset geographical scope but do not meet the preset time period, and the search times correspond to A search 5 times. B searches 7 times, D, E are search words that meet the preset time period but do not meet the preset geographical range. The search times correspond to D search 4 times, E search 10 times, and C and F are the presets. The time period is in accordance with the search term of the preset geographical area, and the corresponding search times are C 2 times and F 4 times, then the 6 popular hot search words in the second search word set are high to low. The sorting is: FCEBAD, in which, although the search times of C and F are small, but the corresponding matching conditions are more consistent, the corresponding sorting order is higher than the other four popular hot search terms.

It should be noted that, in the sorting process in the first search word set, the above-mentioned sorting method also applies to the sorting based on the number of search times.

Further, the public hot search term in the second search term set may also perform a filtering operation associated with the user, for example, setting a price interval according to the unit price level of the user history ordering meal, and filtering the popular hot search term by the price interval.

205. Sort the preset number of to-be-pushed search words from the first search word set and the second search word set respectively according to the heat ranking of the search words.

The popularity ranking of the search words is embodied in the first search word set as a weighted order of the number of executions of the user and the value of the similarity; in the second set of search words, the ranking of the search times is embodied. Accordingly, the specific process of screening the search terms to be pushed includes:

First, deleting the search term that is included in the second search term set and that is repeated in the first search term set.

Second, the search term comprehensive search times and the similarity values in the first search word set are sorted in descending order.

Third, the search words in the second search term set are sorted in descending order of the number of searches.

Fourth, a preset number of search words are extracted from the first search word set and the second search word set in descending order according to the descending order as the to-be-pushed search words.

Wherein, the second step and the third step are not differentiated in the execution order, and the execution of the first step ensures that the subsequent extraction of the search term to be pushed does not extract from the two sets. The same search term. According to the deletion of the repeated search words in the second set of search words, it can be seen that, in this embodiment, the priority level of the first search word set is higher than the second search word set, that is, the screening priority of the search word to be pushed is based on the user. The preferences are chosen.

In addition, when sorting the search words in the first search word set, the search times are mainly for the search words corresponding to the user's historical behavior data, and the similarity values are mainly for other search words calculated by the similarity degree, and A search term having a search number in a sort is higher than a search word having a similarity value, for example, there are six search words A, B, C, D, E, and F, wherein A and B are historical behaviors. For the search words corresponding to the data, the corresponding search times are A: 5 times, B: 10 times; C and D are similar search words of A, the similarity is C: 90%, D: 95%, E, F For a similar search term of B, the similarity is E: 95%, F: 90%, then after sorting, the obtained order is BAEFDC.

The ordering in the second set of search words is described in the above step 204. The ordering in this step can be obtained based on the ordering in 204, and therefore will not be described again in this step.

For the extracted preset number of search terms to be pushed, it is set according to the needs of the specific application, for example, the preset quantity is determined according to the number of search words that can be displayed in the display page. Moreover, the preset number of values may be the same quantity value for the first search word set and the second search word set, or may be different quantity values.

206. Push the search term to be pushed to the corresponding user.

This step is for the display interface of the ordering application, that is, the display area in which the search term is set in the search area. Generally, the page is a secondary page of the application, that is, after the user clicks on the search area in the first page to determine the execution of the search operation, the application jump value search page is provided, except that the input area with the search content is provided in the page. A corresponding search term display area is also provided. Further, the search terms may also be displayed in a column according to a preset classification.

For the above-mentioned display page, in the process of pushing the search word to be pushed in this embodiment, the following steps are specifically included:

First, mark the classification identifier for the search term to be pushed. The classification identifier is mainly used to distinguish the search dimension of the search term and the location of the page display. For example, the search dimension includes a store, a dish, a popular, etc., the store is a word to be pushed to search for the name of the store, and the dish is to be pushed for search. The word is the word of the dish name, and the popular is the most popular search word in the first search word set and the second search word set, and these search words are not limited to a shop or a dish.

Secondly, according to the classification identifier, the to-be-pushed search term is correspondingly pushed to the category push column of the user's display page. The classification push bar is based on the preset classification, and the display column is set in the page, and the search dimension displayed in the column is displayed in different display columns to be viewed by the user classification.

According to the effect of the search word pushed on the user side, the display effect of the search word shown in FIG. 3B in the application page can be seen. The page shown in FIG. 3B is a search page that is jumped after clicking the search area in the first page of the ordering application, in which more detailed search information can be displayed, including a search input bar, a search word push bar, and the like. As shown in FIG. 3B, below the search input bar is a category push bar, wherein "hot", "shop", "dish" are the classification identifiers of the search words, and are displayed according to the tag names marked in the search words. Corresponding push bar. In FIG. 3A, the first page of the ordering application is displayed, that is, the upper page of FIG. 3B, wherein the search bar is framed, that is, the search area in the above step, because a large amount of display is needed in the home page. Information, therefore, in FIG. 3A, the search area only displays the search bar, and the search word push bar is not displayed, then the search words based on the above-mentioned searched words will be displayed in the search bar in a grayscale manner, and need to be explained. Because of the limited display range, the search terms selected for the poem will be further filtered according to the order of sorting to determine the display content in the search bar.

The above is a specific manner for being able to extract a sufficient number of search words from the first set of search words and the second set of search words, and to push the search words when a sufficient number of search words cannot be obtained according to the above manner. When the number is lower than a preset threshold, this situation usually occurs when a new user uses a subscription application in a relatively remote area or during a non-meal time. This situation is difficult to occur, and it is a kind of In a special case, in this case, the application can display the specific content in the search word pushing column of the page, and the search term specified by the system will be promoted. The manner of specifying the search words is not specifically limited in this embodiment.

It can be seen from the above description of the search term pushing method shown in FIG. 2 that this embodiment is a more detailed description of how to obtain the first search word set and the second search word with respect to the push method shown in FIG. 1 . The search words in the collection, in particular, the user history behavior is filtered by the acquired user identity information, location information, and current time information, so that the search terms in the first search term set are closer to the user's preferences, and can be more consistent. The user's current needs, and the location information and current time information are also used in the second set of search terms to filter the popular hot search words, which can provide users with more accurate information in the order application to ensure that the user is using these Search words can get more effective information when querying, thus ensuring the validity of user queries and improving the efficiency of user queries. In addition, by combining with the display page, the embodiment of the present disclosure also provides the user with a variety of search word display manners, which is more convenient for the user to view, and also sets a countermeasure scheme for some special situations, such as insufficient number of search words pushed. The search term is specified by the system, thereby ensuring that the embodiment of the present disclosure can provide the user with sufficient search terms in accordance with their needs in the ordering application.

Through the description of the foregoing embodiment, it can be found that when pushing the search word to the user, the embodiment of the present disclosure not only adds the limited dimension of time and space, but also more importantly adds the user's own preference, that is, through the first search term. The collection is for the user to push the search term that matches the historical operation of the user. Therefore, the embodiment of the present disclosure further describes how to obtain the search term in the first search set by using the following embodiments for the step 203 in the foregoing embodiment. Specifically, as shown in Figure 4, including:

301. Obtain historical behavior data of the user according to the identity information of the user.

Wherein, the historical behavior data for the user in this step may refer to all the behavior data executed by the application recorded in the order application before the user performs this operation. The way to obtain historical behavior data can be distinguished by different dimensions, that is, it can be obtained by a specified quantity or by a specified time.

302. Filter, by using a preset rule, the object name of the user performing the operation from the historical behavior data.

The preset rule in this step specifically includes a quantity rule and a time rule.

Specifically, the quantity rule is: first obtaining a preset number of historical click behaviors closest to the current time, and then extracting object names of the click objects corresponding to the historical click behaviors, wherein the historical click behavior is included in the first page or the secondary list. The operation of clicking the target object in the corresponding page, for example, the user clicks on the specific store operation in the first page of the ordering application, or by clicking the secondary list option in the first page, such as dining, fruit, etc., the application will be the user. The page corresponding to the second-level list is displayed. For the catering, the specific catering-providing store is displayed in the form of a list, and the user clicks on the specific store operation performed on the page. The click object corresponding to these click operations is a store, and the object name is the corresponding store name. Of course, the above click operation may also include clicking on a specific dish, and the object name corresponds to the dish name. For the obtained multiple object names, the heat ranking in the first search set is based on the time distance of the corresponding historical click behavior from the current time, and the closer the heat is, the higher the heat is.

The time rule is: obtaining the object name of the user operation object in the historical behavior data of the user in the preset time period. The content operated by the user is not limited to operations such as clicking, accessing or searching, and the ordering of the obtained object names in the first search set is mainly based on the time and number of times the user performs the operation on the operation object. The sort weight is configured, that is, the closer the weight of the operation time is to the current time, the larger the weight value of the configuration of the same operation object is. The weight of the configuration is larger according to the configured sort weight. Sort. When there is a large amount of historical behavior data in the time rule and need to be filtered, the preset number of object names can be extracted according to the ordering weight from high to low.

It should be noted that the foregoing quantity rule and time rule may be used alone or in combination to obtain multiple object names. The resulting object name can be considered as a specific vocabulary that represents the user's preference for the use of the ordering application.

303. Determine a vocabulary whose similarity with the object name reaches a threshold as a user hot search term.

The similarity calculation method provided in this step is the cosine similarity calculation. To perform the cosine similarity calculation on the object name, the object name needs to be vectorized first, and the vectorized representation needs to obtain the object name. Identification information, wherein the identification information is an identifier for performing preset dimension classification on the operation object, and the preset dimension is a basis for vectorization representation, for example, the preset dimension may include Chinese food, western food, Korean barbecue in the dining Japanese cuisine, fast food, local dishes, etc. If the object name is a hamburger, the labeling information is Western food and fast food.

Secondly, based on the vectorization of the object name, the similarity value of the object name and other object names is calculated by the cosine similarity, wherein the other object names refer to all the records recorded in the ordering application are not filtered in step 302. The object name obtained. By calculation, other object names whose similarity values are greater than the threshold value are retained, so that the purpose of vocabulary expansion based on the user's operation record can be achieved, and the expansion requirement from one word to multiple words can be realized. And the way to expand by the similarity is that the number of other object names obtained can be adjusted by setting the threshold, that is, the number of extensions for the vocabulary is controllable in this step.

Finally, the obtained other object names and the object names selected in step 302 are set as user hot search words, and these user hot search words constitute the first search word set.

Further, on the basis of the foregoing steps, in order to optimize the search words in the first search term set, the steps further include:

304. Filter the user hot search words by using preset filtering conditions.

The preset filtering condition includes at least one of a store classification, a current business status, a distance from the user, a quality evaluation, and a sales quantity. Specific to the ordering application, the store classification refers to the business scope of the store, which should be the catering category; the current business status is to obtain the store that is currently in normal business; the distance from the user is the store whose filtering distance is greater than the threshold For example, a shop with a filtering distance of 5km or more; a quality evaluation is to filter a shop with a rating of excellent or good; and a sales quantity is a shop that filters sales within a specified time period without reaching a threshold. These filters can be better filtered. User hot search words in the first search word set, and these search words can also provide users with more optimized search results.

305. Simplify the text processing of the user hot search words.

Among them, the main purpose of text simplification processing is to extract the core words in the object name to obtain a more simplified, and at the same time suitable for the user's hot search. Generally, the display requirement for the user's hot search term is to display a simple vocabulary or a phrase, and the extracted object name is often mixed with unnecessary content, such as punctuation marks, modifiers, and the like. In this regard, the text simplification processing is to delete the unnecessary content, and the specific deletion method can be deleted according to the relevant rules of the natural language processing, and the details are not elaborated in this step.

The search word screening method shown in FIG. 4 above is mainly for screening the first search word set to match the user hot search words, and by further classifying the user historical behavior data, the selected historical behavior data is selected and extracted. The corresponding object name is further expanded based on the object name, and the similarity calculation is used to obtain more object names that may be of interest to the user, thereby realizing reasonable extension of the user's favorite content and ensuring the search term in the first search word set. It can cover the user's possible preferences within a certain range and increase the accuracy of predicting user needs.

In combination with the content of the foregoing search word pushing method, the specific steps may be simplified by the flowchart shown in FIG. 5, as shown in FIG. 5, wherein the historical behavior data is acquired based on the quantity rule, and the history is acquired based on the time rule. The behavior data and the popular search behavior are based on the user identification information, the location information, and the current time information acquired by the application when the user uses the subscription application, and the data information extracted from the user historical behavior information and the popular search behavior. For the user history behavior information, the embodiment of the present disclosure limits the user's personal preference to the extension by using the corresponding name corresponding to the behavior information, such as similar recommendation, filtering processing, and text processing, and finally obtains the first search word set, specifically Refer to the detailed description in Figure 4. For the public search behavior, the public search term is further filtered according to the behavior to form a second search term set. For details, refer to the detailed description of step 204 in FIG. 2 . After the first search word set and the second search word set are obtained, the search words therein are subjected to a de-reordering process, that is, the detailed content in step 205, and the search words pushed to the user front end for display are selected according to the sorting. . In the process, when the number of searched keywords is insufficient, a system-specified search term is introduced to supplement the search words pushed to the user. In this way, the search words with higher probability to meet the user's needs can be pushed for the user, so that the user directly performs the search query through the pushed search words, thereby improving the frequency of the user's use of the push search.

Through the application of the above method embodiment, the inventor has greatly improved the click rate and the order conversion rate of the recommended search words in the actual application test, wherein the click rate is increased from 5% to 13%, and the order conversion rate is improved by 3% to 5%. It can be seen that applying the embodiment of the present disclosure to an actual product can effectively improve the user experience and promote the success rate of the platform transaction.

Further, as an implementation of the foregoing method embodiment, an embodiment of the present disclosure provides a search term pushing device, where the device is disposed in a smart terminal used by a user, and the device embodiment corresponds to the foregoing method embodiment, and is convenient for reading. The embodiment of the present invention is not described in detail in the foregoing method embodiments, but it should be clear that the device in this embodiment can implement all the contents in the foregoing method embodiments. Specifically, as shown in FIG. 6, the device includes: an obtaining module 41, a first extracting module 42, a second extracting module 43, a screening module 44, and a pushing module 45, where

The obtaining module 41 is configured to obtain the identity information, the location information, and the current time information of the user. In the information source obtained by the specific acquiring module, the identity information is mainly based on the login mode of the user using the application, that is, the registered user login or the guest login. The location information and the current time information are mainly based on the information data provided by the smart terminal used by the user. The obtaining module acquires identity information, location information, and current time information of the user when it is determined that the user opens and enters the application.

The first extraction module 42 is configured to extract a user search term corresponding to the user from the user historical behavior data corresponding to the identity information acquired by the acquiring module 41, to obtain a first search term set, where the historical behavior The data mainly reflects the user's own favorite features, which are the specific operations that the user has performed in the application. It can be seen that the search words in the first set of search words tend to recommend the search words preferred by the user.

The second extraction module 43 is configured to acquire the public hot search term corresponding to the time information acquired by the obtaining module 41, and obtain a second search term set, where the public hot search term is used by the mass user. The search term used in the search operation is performed at the location and time corresponding to the time information, and the public user herein may also include the user. It can be seen that the search words in the second set of search words tend to recommend popular search terms.

The screening module 44 is configured to filter the preset quantity from the first search word set extracted by the first extraction module 42 and the second search word set extracted by the second extraction module 43 according to the heat ranking of the search words. The search term to be pushed, wherein the heat ranking is to sort the search words in each set, and the user prefers or uses frequently as the preferred search word to be pushed. The specific manner of the hot sorting is not limited to the number of times the user performs or the time of execution, or the manner of comprehensive sorting by using multiple dimensions.

The pushing module 45 is configured to push the to-be-pushed search word selected by the screening module 44 to the user, and display it through an application interface used by the user.

Further, as shown in FIG. 7, the first extraction module 42 includes:

The obtaining unit 421 is configured to acquire historical behavior data of the user according to the identity information, where the historical behavior data includes data generated by at least one of a click behavior, a search behavior, and an access behavior, and The content of the behavioral data may also include other activities such as shopping carts, collections, etc., to expand and enrich the user's preferences.

The filtering unit 422 is configured to filter, by using the preset rule, the object name of the user performing the operation from the historical behavior data acquired by the acquiring unit 421, where the preset rule refers to a specific rule for extracting the object name of the user operation object, Different preset rules have different object names.

a determining unit 423, configured to determine a vocabulary whose similarity with the object name selected by the filtering unit 422 reaches a threshold value as a user hot search term, wherein calculating the similarity may adopt a plurality of calculating manners, thereby determining the The range of user preferences expands more hot search terms.

Further, as shown in FIG. 7, the screening unit 422 includes:

The first screening sub-unit 4221 is configured to acquire a preset number of historical click behaviors that are closest to the current time, and extract the historical click behavior corresponding to the object name of the click object, where the historical click behavior is included in the page corresponding to the first page or the second level list. In the operation of clicking the target object, it can be seen that the first filter sub-unit adopts a quantity rule, that is, acquires a preset number of object names.

a second screening subunit 4222, configured to acquire an object name of the user operation object in the historical behavior data of the user in the preset time period, where the difference from the first screening subunit is different from the preset rule used, The second screening sub-unit adopts a time rule, that is, obtains the object names corresponding to all historical behavior data in a certain period of time, and the number thereof is not determined.

Further, the second screening sub-unit 4222 is further configured to: configure a sorting weight based on a time and a number of times the user operates the operation object, and sort the object names of the operation object according to the sorting weight, according to the sorting The weights are extracted from the highest to the lowest order in the preset number of object names.

Further, as shown in FIG. 7, the determining unit 423 includes:

The obtaining sub-unit 4231 is configured to obtain the identifier information of the object name, where the identifier information is an identifier for performing preset dimension classification on the operation object, and the preset dimension is used for vectorizing the object name. The basis, and then the similarity calculation.

The calculating sub-unit 4232 is configured to calculate, by cosine similarity calculation, another object name whose similarity with the identification information acquired by the acquiring sub-unit 4231 is greater than a threshold.

The determining subunit 4233 is configured to determine the object name and the vocabulary corresponding to the other object names obtained by the calculating subunit 4232 as the user hot search term, thereby forming the first search word set.

Further, as shown in FIG. 7, the first extraction module 42 further includes:

a filtering unit 424, configured to: after the determining unit 423 determines a vocabulary whose degree of similarity with the object name reaches a threshold as a user hot search term, filter the user hot search term by using a preset filtering condition, where The filtering conditions include at least one of a store classification, a current business status, a distance from the user, a quality evaluation, and a sales quantity.

The processing unit 425 is configured to perform text simplification processing on the user hot search term determined by the determining unit 423 and/or the user hot search result filtered by the filtering unit 424, where the text simplification processing is used to extract the user hot search term The core word in the middle. The general expression of the core word is a simple vocabulary or phrase, which does not contain punctuation or modifiers.

Further, as shown in FIG. 7, the second extraction module 43 includes:

The first obtaining unit 431 is configured to acquire the popular hot search term in the preset geographical area by using the current location information. The location information is mainly based on the location acquired by the obtaining module 41, and determines the preset geographical extent to which the location belongs, and obtains the popular hot search term within the preset geographical scope.

The second obtaining unit 432 is configured to acquire the popular hot search term in the preset time period by using the current time information. The time information is also based on the current time acquired by the obtaining module 41, and determines the preset time period to which the time belongs, and obtains the popular hot search term within the preset time period.

The sorting unit 433 is configured to: the first obtaining unit 431 acquires the popular hot search term in the preset area range, and the second obtaining unit 432 acquires the popular hot search term in the preset time period in descending order according to the number of searches.

The extracting unit 434 is configured to extract a preset number of popular hot search words in a descending order according to the sorting unit 433.

Further, as shown in FIG. 7, the device further includes:

The determining module 46 is configured to determine whether the current user has user historical behavior data within a preset geographical area and a preset time period, where the preset geographic range is associated with the location information, and the preset time period and the location The current time information is associated. The judging module 46 is mainly configured to set a further screening condition for the user hot search in the first search term set, so that the obtained historical behavior data is more targeted, and has more reference value to the current behavior of the user.

The first extraction module 42 is further configured to: when the determining module 46 determines that the user historical behavior data exists, perform extracting a hot search term corresponding to the user from the user historical behavior data corresponding to the identity information, Get the first set of search terms.

Further, as shown in FIG. 7, the screening module 44 includes:

The deleting unit 441 is configured to delete the search term that is included in the second search term set and is repeated in the first search term set. Avoid extracting the same search term from the second set of search terms as in the first set of search terms.

The sorting unit 442 is configured to sort the search word comprehensive search times and the similarity values in the first search word set in descending order. The number of searches is the number of times the search term corresponding to the user's historical behavior data is generated by the user, and the value of the similarity refers to the value corresponding to the search term calculated by the similarity, and the number of searches in the sorting process. The priority is higher than the value of similarity.

The sorting unit 442 is further configured to sort the search words in the second set of search words in descending order of the number of searches.

The extracting unit 443 is configured to extract, from the first search term set and the second search term set, a preset number of search words in a descending order according to the sorting unit 442, as a to-be-pushed search. word.

Further, as shown in FIG. 7, the push module 45 includes:

The marking unit 451 is configured to mark a classification identifier for the to-be-pushed search term, where the classification identifier is used to distinguish the search dimension of the search term and the location of the page display.

The pushing unit 452 is configured to push the to-be-pushed search term correspondingly to the category pushing column of the user's display page according to the classification identifier marked by the marking unit 451.

Further, the pushing module 45 is further configured to: when the number of search words to be pushed selected by the screening module 44 is lower than a threshold, push the search term specified by the system to the user, so that the user cannot be based on the user Personal preference and when pushing suitable search terms to users through time and location information, it is ensured that there are user-clickable search terms in the search term display column of the display page by pushing the specified search words.

Further, an embodiment of the present disclosure further provides a search term pushing device, the device comprising a processor and a memory, wherein the memory is configured to store one or more computer instructions, and the one or more computer instructions are The processor executes the steps of implementing the search term pushing method shown in FIG. 1, FIG. 2 or FIG. 4 described above.

Further, an embodiment of the present disclosure further provides a computer readable storage medium having stored thereon computer instructions, wherein the searching shown in FIG. 1, FIG. 2 or FIG. 4 is implemented when the computer instructions are executed by a processor. The steps of the word push method.

In summary, a search word pushing method, device, and terminal used in the embodiments of the present disclosure are applied to a subscription application, and the personalized search term is pushed to the user to meet the user's search requirement. Specifically, the user searches for search words from multiple dimensions, and filters the search words for the user from the dimensions of time, location, and historical behavior of the user, and saves the obtained filtered search words in the first search word set and the second search respectively. In the word set, the difference between the two is that the search words in the first search word set are closer to the user's preference, and are the search words obtained by analyzing the historical behavior of the same user and expanding according to the historical behavior. The second set of search words is filtered based on the search behavior of the mass user, wherein the search term can be regarded as a more comprehensive reference information provided to the user on the basis of the first set of search words, through a combination of two sets. Provide users with more accurate search terms, thereby increasing the probability that the pushed search terms will be clicked by the user. For the user, the push method of the search word can quickly find the content to be searched from the searched words, and achieve the purpose of quick query by one click, thereby greatly improving the user experience. In addition, some special cases are also considered in the embodiment of the present disclosure. When the number of search words pushed is insufficient, the search words specified by the synchronization collection system may be pushed to the user, so that the user can conveniently query through the pushed search words without Query by means of information entry.

The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, ie may be located A place, or it can be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the embodiment. Those of ordinary skill in the art can understand and implement without deliberate labor.

Through the description of the above embodiments, those skilled in the art can clearly understand that the various embodiments can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware. Based on such understanding, the above-described technical solutions may be embodied in the form of software products in essence or in the form of software products, which may be stored in a computer readable storage medium such as ROM/RAM, magnetic Discs, optical discs, etc., include instructions for causing a computer device (which may be a personal computer, server, or network device, etc.) to perform the methods described in various embodiments or portions of the embodiments.

It should be noted that the above embodiments are only for explaining the technical solutions of the present disclosure, and are not intended to be limiting; although the present disclosure has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that The technical solutions described in the foregoing embodiments are modified, or the equivalents of the technical features are replaced by the equivalents. The modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present disclosure.

Claims

A search word pushing method, including:

Obtain the user's identity information, location information, and current time information;

Extracting a hot search term corresponding to the user from the user historical behavior data corresponding to the identity information, to obtain a first search term set;

Obtaining a public search term corresponding to the location information and the time information, to obtain a second search term set, wherein the public hot search term is used by a mass user to perform a search operation at a location and time corresponding to the location information and the time information. Search term;

Sorting a preset number of to-be-pushed search words from the first search word set and the second search word set respectively according to the heat ranking of the search words;

Pushing the to-be pushed search term to the user.
The method of claim 1, wherein extracting the user hot search term corresponding to the user from the user historical behavior data corresponding to the identity information comprises:

Obtaining historical behavior data of the user according to the identity information, where the historical behavior data includes data generated by at least one of a click behavior, a search behavior, and an access behavior;

Filtering, by using a preset rule, the object name of the user performing the operation from the historical behavior data;

A vocabulary whose similarity with the object name reaches a threshold is determined as a user hot search term.
The method according to claim 2, wherein the screening of the object name of the user performing the operation from the historical behavior data by using a preset rule comprises:

Obtaining a historical click behavior of a preset number that is closest to the current time, and extracting an object name of the click object corresponding to the historical click behavior, where the historical click behavior includes an operation of clicking the target object in the page corresponding to the first page or the second level list; and/ or

Obtaining an object name of the user operation object in the historical behavior data of the user in the preset time period.
The method according to claim 3, wherein acquiring the object name of the user operation object in the historical behavior data of the user within the preset time period comprises:

Configuring a ranking weight based on the time and number of times the user operates the operation object;

Sorting object names of the operation object according to the sort weight;

The preset number of object names are extracted in descending order of the sort weights.
The method according to claim 2, wherein determining the vocabulary whose degree of similarity with the object name reaches a threshold as the user hot search includes:

Obtaining identification information of the object name, where the identifier information is an identifier for performing preset dimension classification on the operation object;

Calculating, by cosine similarity, another object name whose similarity with the identification information is greater than a threshold;

The vocabulary corresponding to the object name and other object names is determined as a user hot search term.
The method according to any one of claims 2 to 5, wherein after determining a vocabulary whose degree of similarity with the object name reaches a threshold value as a user hot search term, the method further comprises:

The user hot search is filtered by using a preset filter condition, where the filter condition includes at least one of a store classification, a current business status, a distance from the user, a quality evaluation, and a sales quantity;

Text simplification processing is performed on the user hot search word, and the text simplification processing is used to extract a core word in the user hot search term.
The method according to claim 1, wherein the obtaining the public information of the location information corresponding to the time information comprises:

Using the current location information to obtain popular hot search terms within a preset geographic area;

Using the current time information to obtain a popular hot search term within a preset time period;

Sorting the popular hot search words in the preset geographical area and the popular hot search words in the preset time period according to the number of searches;

A preset number of popular hot search words are extracted from the high to the bottom according to the descending order.
The method of claim 1 or 7, wherein the method further comprises:

Determining whether the current user has user historical behavior data within a preset geographical area and a preset time range, the preset geographical range being associated with the location information, the preset time period being associated with the current time information ;

If yes, the user hot search term corresponding to the user is extracted from the user historical behavior data corresponding to the identity information, to obtain a first search term set.
The method according to claim 1, wherein the screening of the preset number of to-be-pushed search words from the first search word set and the second search word set respectively according to the heat ranking of the search words comprises:

Deleting a search term that is included in the second search term set and that is repeated in the first search term set;

Sorting the search word comprehensive search times and the similarity values in the first search word set in descending order;

Sorting the search words in the second set of search words in descending order of the number of searches;

And extracting, from the first search term set and the second search term set, a preset number of search words as a to-be-pushed search term according to the descending order according to the descending order.
The method of claim 1 or 9, wherein pushing the search term to be pushed to the user comprises:

Marking a classification identifier for the to-be-pushed search term, the classification identifier is used to distinguish a search dimension of the search term and a location of the page display;

And correspondingly pushing the to-be-pushed search term to the category push column of the user's display page according to the classification identifier.
The method of any of claims 1-5, 7 and 9, wherein the method further comprises:

When the number of search terms to be pushed is below a threshold, the search term specified by the system is pushed to the user.
A search word pushing device comprising:

An obtaining module, configured to acquire identity information, location information, and current time information of the user;

a first extraction module, configured to extract, from the user historical behavior data corresponding to the identity information acquired by the acquiring module, a user hot search term corresponding to the user, to obtain a first search term set;

a second extraction module, configured to acquire a public hot search term corresponding to the location information acquired by the acquiring module, and obtain a second search term set, where the public hot search term is used by the mass user in the location information and time The search term used by the search operation for the location and time corresponding to the information;

a screening module, configured to filter a preset number of to-be-pushed from the first search word set extracted by the first extraction module and the second search word set extracted by the second extraction module, respectively, according to the heat ranking of the search words Search word

And a pushing module, configured to push the to-be-pushed search word selected by the screening module to the user.
The apparatus of claim 12, wherein the first extraction module comprises:

An acquiring unit, configured to acquire historical behavior data of the user according to the identity information, where the historical behavior data includes data generated by at least one of a click behavior, a search behavior, and an access behavior;

a screening unit, configured to filter, by using a preset rule, the object name of the user performing the operation from the historical behavior data acquired by the acquiring unit;

And a determining unit, configured to determine a vocabulary whose similarity with the object name selected by the screening unit reaches a threshold as a user hot search term.
The apparatus of claim 13 wherein said screening unit comprises:

The first screening sub-unit is configured to acquire a preset number of historical click behaviors that are closest to the current time, and extract the historical click behavior corresponding to the object name of the click object, where the historical click behavior is included in the page corresponding to the first page or the second level list. Click on the action of the target object;

And a second screening subunit, configured to acquire an object name of the user operation object in the historical behavior data of the user in the preset time period.
The apparatus according to claim 14, wherein the second screening subunit is specifically configured to configure an ordering weight based on a time and a number of times the user operates the operation object, and the object of the operation object according to the sorting weight The names are sorted, and the preset number of object names are extracted in descending order of the sort weights.
The apparatus of claim 13, wherein the determining unit comprises:

Obtaining a subunit, configured to obtain identifier information of the object name, where the identifier information is an identifier for performing preset dimension classification on the operation object;

a calculation subunit, configured to calculate, by cosine similarity calculation, another object name whose similarity with the identification information acquired by the acquisition subunit is greater than a threshold;

Determining a subunit, configured to determine the object name and a vocabulary corresponding to the other object name obtained by the computing subunit as a user hot search term.
The apparatus according to any one of claims 13-16, wherein the first extraction module further comprises:

a filtering unit, configured to: after the determining unit determines a vocabulary whose degree of similarity with the object name reaches a threshold as a user hot search term, filter the user hot search term by using a preset filtering condition, where the filtering The condition includes at least one of a store classification, a current business status, a distance from the user, a quality evaluation, and a sales quantity;

And a processing unit, configured to perform text simplification processing on the user hot search term determined by the determining unit, where the text simplification process is used to extract a core word in the user hot search term.
The apparatus of claim 12, wherein the second extraction module comprises:

a first obtaining unit, configured to use the current location information to obtain a popular hot search term in a preset geographical area;

a second obtaining unit, configured to use the current time information to obtain a popular hot search term in the preset time period;

a sorting unit, configured to: the first obtaining unit acquires the popular hot search term in the preset geographical area and the public hot search term in the preset time period of the second obtaining unit to perform descending order according to the number of searches;

And an extracting unit, configured to extract a preset number of popular hot search words in a descending order according to the sorting unit.
The device according to claim 12 or 18, wherein the device further comprises:

a judging module, configured to determine whether the current user has user historical behavior data in a preset geographical area and a preset time period, where the preset geographical range is associated with the location information, the preset time period and the Current time information is associated;

The first extraction module is further configured to: when the determining module determines that the user historical behavior data exists, perform extracting a hot search term corresponding to the user from the user historical behavior data corresponding to the identity information, to obtain a A collection of search words.
The apparatus of claim 12 wherein said screening module comprises:

a deleting unit, configured to delete a search term that is included in the second search term set and that is repeated in the first search term set;

a sorting unit, configured to sort the search word comprehensive search times and the similarity values in the first search word set in descending order;

The sorting unit is further configured to sort the search words in the second set of search words in descending order of the number of searches;

And an extracting unit, configured to extract, from the first search term set and the second search term set, a preset number of search words as a to-be-pushed search term, in descending order of the sorting unit according to a descending order.
The apparatus according to claim 12 or 20, wherein the push module comprises:

a marking unit, configured to mark a classification identifier for the to-be-pushed search term, where the classification identifier is used to distinguish a search dimension of the search term and a location of the page display;

And a pushing unit, configured to push the to-be-pushed search term correspondingly to the category pushing column of the user's display page according to the classification identifier of the marking unit tag.
The apparatus according to any one of claims 12-16, 18, 20, wherein the pushing module is further configured to: when the number of search words to be pushed selected by the screening module is lower than a threshold, the pushing system The specified search term to the user.
A search term push terminal, wherein the terminal comprises a processor and a memory, wherein the memory is for storing one or more computer instructions, the one or more computer instructions being executed by the processor to implement a right The step of the search term pushing method of any one of 1 to 11.
A computer readable storage medium having stored thereon computer instructions, wherein the computer instructions are executed by a processor to perform the steps of the search term pushing method of any one of claims 1 to 11.