CN112100480A

CN112100480A - Search method, device, equipment and storage medium

Info

Publication number: CN112100480A
Application number: CN202010969712.XA
Authority: CN
Inventors: 贺小林; 范淼
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2020-09-15
Filing date: 2020-09-15
Publication date: 2020-12-18
Anticipated expiration: 2040-09-15
Also published as: CN112100480B

Abstract

The application discloses a searching method, a searching device, searching equipment and a storage medium, and relates to the field of intelligent searching. The specific implementation scheme is as follows: acquiring a search vocabulary; obtaining a prediction vocabulary corresponding to the search vocabulary, wherein the prediction vocabulary comprises the search vocabulary; acquiring a first search result set corresponding to the search vocabulary and a second search result set corresponding to the prediction vocabulary; and determining a target search result according to the first search result set and the second search result set, and displaying the target search result. The effective target search result can be obtained on the premise that the user inputs limited search words, and the search is more accurate and efficient.

Description

Search method, device, equipment and storage medium

Technical Field

The present application relates to intelligent search in data processing, and in particular, to a search method, apparatus, device, and storage medium.

Background

Generally, in a search engine, a user inputs a segment of a keyword to be queried, and the search engine provides a corresponding search suggestion according to the segment of the keyword, and then executes a search process according to the search suggestion selected by the user to obtain a search result.

In the map search engine, generally, when a user inputs a segment of a keyword to be queried, the map search engine determines and displays a search result according to the segment of the keyword, and the map search engine does not select a search suggestion and performs a search process according to the search suggestion selected by the user.

Since the segments of the keywords contain fewer text features and have ambiguous semantics, determining the search result directly according to the segments of the keywords causes a large difference between the search result required by the user and the actual search result, and thus, the user cannot obtain an effective search result on the premise of inputting fewer segments of the keywords.

Disclosure of Invention

The application provides a searching method, a searching device, searching equipment and a storage medium.

According to a first aspect of the present application, there is provided a search method comprising:

acquiring a search vocabulary;

acquiring a prediction vocabulary corresponding to the search vocabulary, wherein the prediction vocabulary comprises the search vocabulary;

acquiring a first search result set corresponding to search words and a second search result set corresponding to prediction words;

and determining a target search result according to the first search result set and the second search result set, and displaying the target search result.

According to a second aspect of the present application, there is provided a search apparatus comprising:

the first acquisition module is used for acquiring search words;

the second acquisition module is used for acquiring a predicted vocabulary corresponding to the search vocabulary, and the predicted vocabulary comprises the search vocabulary;

the third acquisition module is used for acquiring a first search result set corresponding to the search vocabulary and a second search result set corresponding to the prediction vocabulary;

and the processing module is used for determining a target search result according to the first search result set and the second search result set and displaying the target search result.

According to a third aspect of the present application, there is provided an electronic device comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of the first aspects.

According to a fourth aspect of the present application, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method of any one of the first aspects.

According to the searching method, the searching device, the searching equipment and the storage medium, firstly, searching words are obtained, the searching words are words input by a user, then, prediction words corresponding to the searching words are obtained, the prediction words are words obtained by reasonably predicting the searching words, the prediction words comprise the searching words, and the prediction words have more text characteristics compared with the searching words; then, a first search result set corresponding to the search words and a second search result set corresponding to the prediction words are obtained, target search results are determined according to the first search result set and the second search result set, and the target search results are displayed. According to the scheme of the embodiment of the application, the predicted words are determined by searching the words, the included text characteristics are increased, the semantics of the words are more clear, then the first search result set and the second search result set are determined according to the searched words and the predicted words, effective target search results can be obtained on the premise that a user inputs limited searched words, and the search is more accurate and efficient.

It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present application, nor do they limit the scope of the present application. Other features of the present application will become apparent from the following description.

Drawings

The drawings are included to provide a better understanding of the present solution and are not intended to limit the present application. Wherein:

FIG. 1 is a schematic diagram of a search engine search;

fig. 2 is a schematic view of an application scenario provided in an embodiment of the present application;

fig. 3 is a schematic flowchart of a searching method according to an embodiment of the present application;

FIG. 4 is a general diagram of a search method provided in an embodiment of the present application;

fig. 5 is a schematic diagram of search results under different location information according to an embodiment of the present application;

fig. 6 is a structural schematic diagram of a GRU model provided in an embodiment of the present application;

FIG. 7 is a schematic structural diagram of a first model provided in an embodiment of the present application;

FIG. 8 is a schematic diagram of obtaining a first sample according to an embodiment of the present disclosure;

FIG. 9 is a schematic diagram of a demand rearrangement provided in an embodiment of the present application;

fig. 10 is a schematic structural diagram of a search apparatus according to an embodiment of the present application;

fig. 11 is a block diagram of an electronic device of a search method according to an embodiment of the present application.

Detailed Description

The following description of the exemplary embodiments of the present application, taken in conjunction with the accompanying drawings, includes various details of the embodiments of the application for the understanding of the same, which are to be considered exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

First, an application scenario of the present application will be described with reference to the drawings.

Search suggestions (sug) are a common product form in a query (query) input process of a search engine, and the search suggestions can be provided in various search types such as general search, vertical search and the like. The main function of sug is to provide some search suggestions to the user and reduce the user input when the user inputs words, and in this type of search, the path for the user to find the required result is:

fragment query-complete query (sug provided) -search results (search engine).

Wherein the segment query is input by the user, and when the user inputs the segment query, the sug provides some suggestions according to the segment query input by the user, wherein the suggestions include one or more complete queries which the user may want to query. When the user clicks one of the complete queries, the search engine executes the search to obtain the search result of the complete query.

Fig. 1 is a schematic diagram of a search engine search, and as shown in fig. 1, on an interface 11 of a terminal device, a user may open a search engine to perform a search. In fig. 1, the search term input by the user is "department", and the "department" at this time is the snippet query.

After acquiring the search vocabulary input by the user, the terminal device may give a corresponding search suggestion, as shown by an interface 12 of the terminal device in fig. 1, where the search suggestion includes "science", "subject two", "science fiction picture", "science and technology university", and the like. The user can click on a search suggestion required by the user, for example, "science and technology university" can be clicked, and then the terminal device can know that the complete query which the user wants to search is "science and technology university" at the moment, and further perform search operation to obtain a related search result of the "science and technology university".

Fig. 1 illustrates a search process of a general search engine, and a search path for some specific searches, for example, in a map search scenario, is different from a general search path, and the search path is generally:

fragment query-search results (search engine).

The step of providing a complete query is omitted with respect to the search path of a normal search. Fig. 2 is a schematic view of an application scenario provided in the embodiment of the present application, and as shown in fig. 2, is a search process in a map scenario.

Under an interface 21 of the terminal device, the interface is an interface of map-related applications, a user inputs a ' department ' word, then the interface 21 directly jumps to an interface 22, and a search result of ' science and technology university ' is displayed, wherein the ' department ' is a section query input by the user, and the search result of the "science and technology university ' is a search result corresponding to the section query. This is the purpose that the map search needs to achieve.

In the map search scenario illustrated in fig. 2, since the segment query input by the user is relatively short, it is difficult to predict the real demand of the user. For example, it is difficult for a user to decide which of "science and technology", "science and technology university", "subject", etc. the user needs to search for "subject". Further, in the map search, the requirement of the user is greatly related to time, space factors and the like, for example, in fig. 2, the search result of "science and technology university" is displayed through the "department" input by the user, and the location information may be determined jointly by integrating "beijing".

In the current searching process, unified sequencing is firstly carried out according to the fragment query input by the user. Specifically, the query is recalled according to the fragment to obtain a vocabulary set related to the fragment query. And then, on the basis of recall, firstly carrying out rough arrangement to pass through a part of vocabularies in the vocabulary set, and then sequencing the rest vocabularies to obtain a sequencing result. On the basis of unified sorting, the sorting result can be finely adjusted according to characteristics such as clicking, impurity identification and the like.

In the above scheme, under the condition that the fragment query is short, the text features contained in the fragment query are very limited, the difference between the text vocabularies obtained through recalling is small, the semantics are not clear enough, so that the user demand points are easily filtered out in the rough ranking stage or are ranked to the back position in the unified ranking process, and an effective search result cannot be obtained.

Based on this, the embodiment of the application provides a search scheme, so as to realize effective positioning of a user demand point under the condition that the fragment query is short, and improve the search efficiency and accuracy. The scheme of the present application will be described below.

Fig. 3 is a schematic flowchart of a search method provided in an embodiment of the present application, and as shown in fig. 3, the method may include:

and S31, acquiring a search word.

The search term may include one or more words, and the search term may be a complete term or a portion of a complete term. The search vocabulary can be input by the user in an input window of the search engine, and then the search vocabulary input by the user is acquired in the background.

For example, if the user wants to search for "science and technology university," the user may enter "subject" in the search engine, where "subject" is the search term and "subject" is a part of a complete term; the user may also enter "science" in the search engine, where "science" is the search term and "science" may exist as a complete term but not identical to "science university" that the user wants to search.

And S32, acquiring a predicted vocabulary corresponding to the search vocabulary, wherein the predicted vocabulary comprises the search vocabulary.

After the search vocabulary input by the user is obtained, the corresponding prediction vocabulary is obtained according to the search vocabulary. In the embodiment of the application, the prediction vocabulary is a vocabulary obtained by reasonable prediction according to the search vocabulary, the prediction vocabulary comprises the search vocabulary, the text in the search vocabulary also exists in the prediction vocabulary, and the sequence of the words in the text in the search vocabulary is not changed with the sequence of the words in the corresponding text in the prediction vocabulary.

For example, when the search term is "science", the corresponding prediction term may be "science" including "science" words; when the search vocabulary is "fun", the corresponding prediction vocabulary may be "casino", which includes the vocabulary "fun", etc.

S33, a first search result set corresponding to the search words and a second search result set corresponding to the prediction words are obtained.

After the search vocabulary is obtained, a corresponding first search result set is determined according to the search vocabulary, namely, the search result is recalled according to the search vocabulary. There are various ways to determine the corresponding first set of search results based on the search terms, including for example, prefix recalls, suffix recalls, and the like. In the example of postfix recall, after a search term is obtained, all search results beginning with the search term may be added to the first set of search results. For example, taking the search term as "department" as an example, related search results including "department ratio", "department two", "department three", "department creation picture", "science and technology board" may be obtained, so as to determine the first search result set.

After the predicted vocabulary is determined, a corresponding second search result set is also determined according to the predicted vocabulary, namely, the search results are recalled according to the predicted vocabulary. In the embodiment of the present application, the manner of recalling the search result according to the predicted vocabulary is similar to the manner of recalling the search result according to the search vocabulary, and various manners including prefix recall, suffix recall, and the like can be adopted. The difference is that the search result range included in the second search result set determined according to the predicted vocabulary is smaller and more targeted because the predicted vocabulary includes richer text features.

And S34, determining a target search result according to the first search result set and the second search result set, and displaying the target search result.

After the first set of search results and the second set of search results are determined, the search results in the first set of search results and the second set of search results may be ranked, and the target search result may be determined and displayed according to the ranking results.

In an embodiment of the present application, determining a target search result combines a first set of search results determined from a search term and a second set of search results determined from a predicted term. Because the predicted vocabulary has more text characteristics, a more accurate and effective target search result can be obtained according to the second search result set determined by the predicted vocabulary. In some cases, however, the predicted vocabulary may not match the user's actual demand points, and thus the first set of search results determined from the original search vocabulary is retained in embodiments of the subject application. The first set of search results and the second set of search results are combined to determine a target search result.

The searching method provided by the embodiment of the application comprises the steps of firstly obtaining a searching vocabulary, wherein the searching vocabulary is a vocabulary input by a user, then obtaining a prediction vocabulary corresponding to the searching vocabulary, wherein the prediction vocabulary is a vocabulary obtained by reasonably predicting the searching vocabulary, the prediction vocabulary comprises the searching vocabulary, and the prediction vocabulary has more text characteristics compared with the searching vocabulary; then, a first search result set corresponding to the search words and a second search result set corresponding to the prediction words are obtained, target search results are determined according to the first search result set and the second search result set, and the target search results are displayed. According to the scheme of the embodiment of the application, the predicted words are determined by searching the words, the included text characteristics are increased, the semantics of the words are more clear, then the first search result set and the second search result set are determined according to the searched words and the predicted words, effective target search results can be obtained on the premise that a user inputs limited searched words, and the search is more accurate and efficient.

The user continuously initiates requests to the sug service in the process of inputting the query (query word) in the map search, and each request is an independent search process for the sug service. For example, if the user wants to find poi "university of people in china", then the input sequence for the user may be: zhong- > China- > Chinese- > … - > university of Chinese people. In any of the requests, the sug returns a result with the user's request poi and is clicked, and the input is terminated, indicating that the user's request is satisfied. When the demand is met, the number of requests received by the server is positioned as the input step length of the user, the input step length is generally in a direct proportion relation with the number of input words, the input step length is the input cost for finding the demand poi by the user, and the smaller the value is, the better the value is.

Generally, the further the user needs poi are ranked, the higher the probability of the user clicking on a sug. The sorting technology based on historical clicking and query prediction is realized on the basis of uniform sorting and requirement rearrangement, and the input step length of a user can be optimized to the maximum extent. In the embodiment of the application, on the basis, the predicted query is obtained through the fragment query input by the user, so that the requirement of the user is predicted under the condition of a short and most incomplete semantic queries, and the user requirement poi is arranged in front of the predicted query.

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

Fig. 4 is a general schematic diagram of a search method provided in the embodiment of the present application, and as shown in fig. 4, the search method includes several main stages of demand recall, rough truncation, uniform sorting, demand rearrangement, and the like, and finally obtains a final sorting result.

Before the demand recall phase, the user first enters a snippet query, i.e., a search term. The demand recall stage is to recall various vocabulary texts related to the fragment query input by the user, and as described above, the demand recall may include a prefix recall, a suffix recall, and the like. For example, when the query of the segment input by the user is "department", the requirement recall may obtain various text vocabularies related to "department", such as "science", "subject two", "subject three", and so on. In practice, the user input fragment query is a dynamic process, and the demand recall is also dynamic. For example, if the user inputs "skill" after "science" is input, the demand recall may obtain various text vocabularies related to "science and technology", such as "science and technology university", "science and technology center", "science and technology institute", and so on.

The number of words and texts recalled by demand is usually large, so after the demand recall, a rough truncation process is carried out, and the process is to roughly screen the text words and filter a part of the text words and texts recalled by demand. The text vocabularies obtained after the coarse truncation are uniformly ordered.

The unified sequencing process is to obtain a predicted query (i.e. a predicted vocabulary) according to an original fragment query (i.e. a search vocabulary), then determine a corresponding first search result set according to the search vocabulary, determine a corresponding second search result set according to the predicted vocabulary, and perform preliminary sequencing according to the first search result set and the second search result set.

The predicted vocabulary is reasonable prediction of the search vocabulary, and the vocabulary which the user may want to search is predicted according to the search vocabulary input by the user, so that the accuracy of the predicted vocabulary has important influence on the accuracy of a subsequent target search result.

In different scenarios, the same search vocabulary may cause different vocabularies of interest to the user and corresponding predicted vocabularies to be different. In the map search scene, the influence of the position information and the time information on the prediction vocabulary is large. In different spatiotemporal scenarios, the needs of the users may be quite different. The search vocabulary is predicted by combining the position information and the time information, and the obtained prediction vocabulary can be more accurate and efficient and is more matched with the actual requirements of users.

Under different location information, for example, when the search word input by the user is "china", if the user is located in beijing, the next possible input by the user is "people", etc.; if the user is located at fertile, the user's next possible inputs are "science", etc. Under different time information, the same search vocabulary and the interested demand points of the user may be different, for example, in the morning, the user may search more areas such as schools, office buildings and the like, and in the evening, the user may search more areas such as cells, parks and the like. This is explained below with reference to fig. 5.

Fig. 5 is a schematic diagram of search results under different location information according to the embodiment of the present application, and as shown in fig. 5, a certain map application is opened on the terminal device 51, where a location displayed on a display map is "beijing". The user inputs "china" in the input window, and the search result of "university of people in china" in beijing is displayed on the map. On the right side of fig. 5, a certain map-like application is opened on the terminal device 52, wherein the displayed position of the map is "fat". The user inputs 'China' in the input window, and the search result of 'China science and technology university' in fertile state is displayed on the map.

As can be seen from the example of fig. 5, even if the user inputs the same search word, the demand points of the user may be completely different due to differences in location information, time information, and the like.

Therefore, in the embodiment of the application, when the search method is applied to a map search engine, in a map search scene, first position information and/or current time are obtained, and prediction words corresponding to search words are obtained according to the position information and/or the current time. The position information may be position information corresponding to a map currently displayed by a map search engine, or position information of a current position of a user, where the current time is a time when the user inputs a search word.

Specifically, the first vocabulary set can be determined based on the location information and the search vocabulary. For example, a third vocabulary set corresponding to the search vocabulary is first obtained, and the vocabulary in the third vocabulary set is the vocabulary searched by using the search vocabulary as the keyword. Then, the matching degree of the words in the third word set and the position information is judged. For example, in fig. 5, when the search vocabulary is "china", both "chinese university" and "chinese science and technology university" are "china" related vocabularies, and both belong to the third vocabulary set. If the position information is Beijing at this time, the matching degree of the Chinese people university and the Beijing is obviously higher, and the matching degree of the Chinese science and technology university and the Beijing is lower. Accordingly, at least one word in the third set of words that matches the location information may be determined to be the first set of words.

The second set of words can also be determined based on the current time and the search words. Specifically, a fourth vocabulary set corresponding to the search vocabulary may be obtained, where the vocabulary in the fourth vocabulary set is obtained by searching using the search vocabulary as a keyword. And then determining at least one word in the fourth word set, which is matched with the current time, as the second word set. For example, when the search vocabulary is "library", the vocabulary searched for by using the vocabulary of "library" as the keyword includes "library", "museum", "hotel", etc., and all the vocabularies related to "library" belong to the fourth vocabulary set. If the current time is nine hours at night, the matching degree of the hotel and the current time is higher, and the matching degree of the library and the museum is lower. Thus, "hotels" in the fourth collection of words may be considered words in the second collection of words.

It should be noted that, in the above embodiment, the position information is "beijing", the current time is "night nine times", and the like are examples, and are respectively used to explain that the first vocabulary set is determined according to the position information, and the second vocabulary set is determined according to the current time. The determination manner of the second vocabulary set is similar to that of the first vocabulary set, and is not described herein again. Based on the first vocabulary set and/or the second vocabulary set, a predicted vocabulary may be determined.

In another possible implementation manner, the search vocabulary, the position information and the current time can be processed through the first model to obtain a prediction vocabulary; the first model is obtained by learning a plurality of groups of first samples, and each first sample comprises a sample search vocabulary, sample position information, a sample time and a sample prediction vocabulary.

In the scheme of the embodiment of the application, prediction needs to be performed according to search words (segment query) input by a user to obtain prediction words (prediction query), and then according to the correlation between the prediction query and the poi, the poi is sorted through a sorting model to determine the sorting position of the poi.

For example, when the query is "Chinese", the requirement poi "Chinese university" is ranked in the first three places (the ranking interval with the highest probability of being seen and clicked by the user), which indicates that the ranking model can predict the requirement poi according to the correlation characteristics between the query and the requirement poi; when the query is Chinese, the requirement poi is ranked after the fifth position, which indicates that the characteristic is not enough to support the ordering model to predict the requirement poi; when query is "medium", the situation is similar. Then, when the query is "medium" or "china", the scheme of the embodiment of the present application predicts a subsequent possible query, the predicted query may be "chinese", and based on the correlation characteristics calculated by the predicted query, the predicted query is added to the ranking model, the ranking position of the demand poi may be advanced, and from the perspective of the input step, the input step of the user is shortened by one step or two steps.

query prediction can be viewed as a language modeling problem, i.e., predicting the next most likely word or phrase given the preceding portion of a sentence. Recurrent Neural Networks (RNNs) are a modeling tool for language modeling problems that can predict any number of words backwards (or forwards). In the embodiment of the application, a gru (gate recovery unit) model in the RNN is used to model the query prediction.

Fig. 6 is a structural schematic diagram of a GRU model provided in an embodiment of the present application, as shown in fig. 6, where an x-sequence represents an input, an O-sequence represents an output, and an h-sequence represents an intermediate state of the model. Corresponding to the x sequence and the O sequence, the characters can be simply regarded as input and output.

E.g. x_t-1Represents "middle", and the word after "middle" is predicted to be "nation" by the GRU model, then O_t-1Represents "nation"; x is the number of_tRepresents "nation", and the word behind "nation" is predicted to be "human" by GRU model, then O_tRepresents "human"; x is the number of_t+1Representing "person", and predicting the character behind "person" as "people" through GRU model, then O_t+1Indicating "people", etc.

In a map search scene, user requirements may be completely different in different space-time scenes, and next inputs may be completely different in the same query segment, for example, in beijing, the next inputs may be "people", and the like, and in fertility combination, the next inputs may be "science", and the like. The conventional GRU model cannot solve the problem that the prediction query is different under different space-time scenes. Based on this, the GRU model can be improved according to the first model in the embodiment of the present application. Fig. 7 is a schematic structural diagram of a first model provided in the embodiment of the present application, and as shown in fig. 7, current time and location information searched by a user are added to the prediction input of each time step of the GRU model, and structures of other parts are unchanged to form the first model.

After the first model is constructed, the first model also needs to be trained, and training samples need to be obtained before training. Fig. 8 is a schematic diagram of obtaining a first sample according to an embodiment of the present application, and as shown in fig. 8, a search operation by a user is displayed on a terminal device 81. After entering "subject", the search engine gives corresponding search suggestions including "science", "subject", "science and technology university" and "science fiction", and the end user clicks "science", so that a set of samples 82 can be obtained according to the search behavior, where the samples 82 include location information "beijing", current time "14: 22", sample search word "subject" and sample prediction word "skill".

Further, when the model training is performed, the current time can be converted into a corresponding time characteristic, and the position information can be converted into a corresponding position characteristic. For example, the time of day may be divided into 8 buckets and encoded, the user's location information may be encoded, and the spatiotemporal information may be combined into a 40-dimensional vector. The first model is trained by adopting mass query (containing space-time information) actually input by a user, and the accuracy of the first model prediction is improved. And predicting one or more steps based on the currently input search vocabulary by using the trained first model to obtain one or more predicted vocabularies.

Fig. 8 illustrates an obtaining method of a set of first samples, and it should be noted that, when obtaining the first samples, the searching operation is not limited to a certain user history, but a large number of searching operations of different user histories can be collected to form the first samples, so as to train the first model. After the training of the first model is completed, a predicted vocabulary can be obtained according to the first model and the search vocabulary. And then determining a first search result set according to the search words and determining a second search result set according to the predicted words. In the embodiment of the application, the GRU model is improved to obtain the first model, and the influence of the position information and the time information on the predicted vocabulary is integrated, so that the obtained predicted vocabulary is more matched with the actual requirement of a user.

As shown in fig. 4, after the first search result set and the second search result set are determined, the words in the first search result set and the second search result set are sorted to obtain a sorting result, and a target search result is determined according to the sorting result.

One possible implementation is to determine the first N search results in the ranked results as the target search results, where N is an integer greater than or equal to 1.

Another possible implementation is to obtain at least one reference search result and then determine a target search result according to the at least one reference search result and the ranking result.

In fig. 4, a portion of the vocabulary that is frequently clicked by the user may be filtered out when the rough truncation is performed. If the user requirements poi are filtered out, then the user requirements poi cannot be ranked in any subsequent ranking. Aiming at the situation, the vocabulary with the historical click hit rate larger than or equal to the preset threshold value in the vocabulary required to be recalled is added into the bypass, and the personalized demand rearrangement based on the historical click rate is added in the demand rearrangement stage after the unified sequencing.

FIG. 9 is a schematic diagram of a requirement rearrangement process provided by an embodiment of the present application, as shown in FIG. 9, illustrating a set 91 obtained after a requirement recall, where the set 91 includes a plurality of words. The vocabulary in the set 91 is then cut through roughly to obtain a set 92 and a set 93, wherein the set 92 is a filtered vocabulary set including "science and technology", "subject two", and "science and technology university". The set 93 is a set of words that are then uniformly ordered, including "science", "subject", and "scientific".

The vocabulary in the set 93 is ranked to obtain a set 94, the set 94 is a ranked set, and the vocabulary in the set is ranked in order of science, subject and scientific creation.

At this point, if the historical click hit rate of "science and technology university" in set 92 exceeds a preset threshold, it indicates that many users have clicked on the poi. Therefore, the vocabulary is added into the set 94 and is comprehensively ranked with the vocabulary in the set 94, so as to obtain a ranked set 95, wherein the vocabulary in the set 95 is ranked in the order of "science and technology", "science and technology university", "science", "subject", and "scientific creation".

After the requirement rearrangement is performed, the first M search results may be obtained from the sorting results, where M is an integer greater than or equal to 1, and P reference search results may be obtained from the at least one reference search result, where P is an integer, and then, the M search results, and the P reference search results are determined as the target search results. In the embodiment of the application, the reference search result is a search result with a historical click hit rate greater than or equal to a preset threshold, and the historical click hit rate can be obtained from the obtained historical click records of multiple users. Since the historical click hit rate of the reference search result is high, it also indicates that it is likely to be the poi required by the user in the search process. However, in some cases, by ranking the search results as described above, some reference search results with higher historical click-hit rates may be excluded or ranked very late, resulting in the target search results being subsequently determined without the reference search results being presented. Therefore, P reference search results can be obtained, and then M search results and P reference search results are determined as target search results, so that the reference search results with higher historical click hit rate are not excluded.

P may be a positive integer or 0. When a reference search result having a historical click hit rate greater than or equal to a preset threshold is included in the ranked results, for example, the top M search results, P may take 0 at this time. And when P is a positive integer, determining a target search result for the P reference search results according to the M search results. When P is 0, the target search result is determined according to the M search results.

The ranking of the reference search result may be located at any position in the ranking result, or a certain ranking rule may be preset.

For example, in the poi clicked by the user history, the poi with the highest click times in the current map area city, the positioning city and other cities are respectively found, then according to the city of the first poi in the ranking result, the poi of the city is found in the reference search result, the poi of the corresponding city in the ranking result is inserted to the second position and is removed, for example, the first poi is the map area city result, and the poi with the highest click times in the map area city in the reference search result is inserted.

Through the scheme, the poi which can avoid user demands to a certain extent is filtered out, or is arranged to a position behind, so that the determination of the follow-up target search result can be more accurate and efficient.

Fig. 10 is a schematic structural diagram of a search apparatus according to an embodiment of the present application, and as shown in fig. 10, the search apparatus includes:

a first obtaining module 101, configured to obtain a search term;

the second obtaining module 102 is configured to obtain a predicted vocabulary corresponding to the search vocabulary, where the predicted vocabulary includes the search vocabulary;

a third obtaining module 103, configured to obtain a first search result set corresponding to the search vocabulary and a second search result set corresponding to the prediction vocabulary;

and the processing module 104 is configured to determine a target search result according to the first search result set and the second search result set, and display the target search result.

In one possible embodiment, the method is applied to a map search engine; the second obtaining module 102 is specifically configured to:

acquiring position information and/or current time, wherein the position information is position information corresponding to a current display map of a map search engine or position information of a current position of a user;

and acquiring a prediction vocabulary corresponding to the search vocabulary according to the position information and/or the current moment.

In a possible implementation, the second obtaining module 102 is specifically configured to:

determining a first vocabulary set according to the position information and the search vocabulary;

determining a second vocabulary set according to the current time and the search vocabulary;

and determining a predicted vocabulary according to the first vocabulary set and/or the second vocabulary set.

acquiring a third vocabulary set corresponding to the search vocabulary, wherein the vocabulary in the third vocabulary set is obtained by searching by taking the search vocabulary as a keyword;

and determining at least one word in the third word set, which is matched with the position information, as the first word set.

acquiring a fourth vocabulary set corresponding to the search vocabulary, wherein the vocabulary in the fourth vocabulary set is obtained by searching by taking the search vocabulary as a keyword;

and determining at least one word in the fourth word set, which is matched with the current time, as the second word set.

processing the search vocabulary, the position information and the current moment through a first model to obtain a prediction vocabulary; the first model is obtained by learning a plurality of groups of first samples, and each group of first samples comprises sample search words, sample position information, sample time and sample prediction words.

In a possible implementation, the processing module 104 is specifically configured to:

sequencing the search results in the first search result set and the second search result set to obtain a sequencing result;

and determining a target search result according to the sorting result.

and determining the first N search results in the sequencing results as target search results, wherein N is an integer greater than or equal to 1.

obtaining at least one reference search result, wherein the historical click hit rate of the reference search result is greater than or equal to a preset threshold value;

and determining a target search result according to the at least one reference search result and the sequencing result.

obtaining the first M search results from the sorting results, wherein M is an integer greater than or equal to 1;

obtaining P reference search results from the at least one reference search result, wherein P is an integer;

and determining the M search results and the P reference search results as target search results.

The search apparatus provided in the embodiment of the present application is used to execute the method embodiments, and the implementation principle and the technical effect are similar, which are not described herein again.

According to an embodiment of the present application, an electronic device and a readable storage medium are also provided.

Fig. 11 is a block diagram of an electronic device of a search method according to an embodiment of the present application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present application that are described and/or claimed herein.

As shown in fig. 11, the electronic apparatus includes: one or more processors 1101, a memory 1102, and interfaces for connecting the various components, including a high speed interface and a low speed interface. The various components are interconnected using different buses and may be mounted on a common motherboard or in other manners as desired. The processor may process instructions for execution within the electronic device, including instructions stored in or on the memory to display graphical information of a GUI on an external input/output apparatus (such as a display device coupled to the interface). In other embodiments, multiple processors and/or multiple buses may be used, along with multiple memories, as desired. Also, multiple electronic devices may be connected, with each device providing portions of the necessary operations (e.g., as a server array, a group of blade servers, or a multi-processor system). In fig. 11, a processor 1101 is taken as an example.

The memory 1102 is a non-transitory computer readable storage medium as provided herein. The memory stores instructions executable by the at least one processor to cause the at least one processor to perform the search method provided herein. The non-transitory computer-readable storage medium of the present application stores computer instructions for causing a computer to perform the search method provided by the present application.

The memory 1102, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs, non-transitory computer executable programs, and modules, such as program instructions/modules corresponding to the search method in the embodiment of the present application (for example, the first obtaining module 101, the second obtaining module 102, the third obtaining module 103, and the processing module 104 shown in fig. 10). The processor 1101 executes various functional applications of the server and data processing, i.e., implements the search method in the above-described method embodiments, by running non-transitory software programs, instructions, and modules stored in the memory 1102.

The memory 1102 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to use of the electronic device of the search method, and the like. Further, the memory 1102 may include high speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory 1102 may optionally include memory located remotely from the processor 1101, which may be connected to the electronic device of the search method via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The electronic device of the search method may further include: an input device 1103 and an output device 1104. The processor 1101, the memory 1102, the input device 1103 and the output device 1104 may be connected by a bus or other means, and are exemplified by the bus 1105 in fig. 11.

The input device 1103 may receive input numeric or character information and generate key signal inputs related to user settings and function control of the electronic apparatus of the search method, such as an input device of a touch screen, a keypad, a mouse, a track pad, a touch pad, a pointer, one or more mouse buttons, a track ball, a joystick, or the like. The output devices 1104 may include a display device, auxiliary lighting devices (e.g., LEDs), tactile feedback devices (e.g., vibrating motors), and the like. The display device may include, but is not limited to, a Liquid Crystal Display (LCD), a Light Emitting Diode (LED) display, and a plasma display. In some implementations, the display device can be a touch screen.

Various implementations of the systems and techniques described here can be realized in digital electronic circuitry, integrated circuitry, application specific ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.

These computer programs (also known as programs, software applications, or code) include machine instructions for a programmable processor, and may be implemented using high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.

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

The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.

The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present application may be executed in parallel, sequentially, or in different orders, and the present invention is not limited thereto as long as the desired results of the technical solutions disclosed in the present application can be achieved.

The above-described embodiments should not be construed as limiting the scope of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A search method, comprising:

acquiring a search vocabulary;

obtaining a prediction vocabulary corresponding to the search vocabulary, wherein the prediction vocabulary comprises the search vocabulary;

acquiring a first search result set corresponding to the search vocabulary and a second search result set corresponding to the prediction vocabulary;

2. The method of claim 1, wherein the method is applied to a map search engine; obtaining a prediction vocabulary corresponding to the search vocabulary, including:

acquiring position information and/or current time, wherein the position information is position information corresponding to a current display map of the map search engine or position information of a current position of a user;

3. The method according to claim 2, wherein obtaining a predicted vocabulary corresponding to the search vocabulary according to the location information and/or the current time comprises:

and determining the predicted vocabulary according to the first vocabulary set and/or the second vocabulary set.

4. The method of claim 3, determining a first set of words from the location information and the search words, comprising:

determining at least one word in the third word set which is matched with the position information as the first word set.

5. The method of claim 3, determining a second set of words based on the current time and the search word, comprising:

and determining at least one word matched with the current time in the fourth word set as the second word set.

6. The method according to claim 2, wherein obtaining a predicted vocabulary corresponding to the search vocabulary according to the location information and/or the current time comprises:

processing the search vocabulary, the position information and the current moment through a first model to obtain the prediction vocabulary; the first model is obtained by learning a plurality of groups of first samples, and each group of first samples comprises sample search words, sample position information, sample time and sample prediction words.

7. The method of any of claims 1-6, determining a target search result from the first set of search results and the second set of search results, comprising:

sorting the search results in the first search result set and the second search result set to obtain a sorting result;

and determining the target search result according to the sorting result.

8. The method of claim 7, determining the target search result from the ranking result, comprising:

and determining the first N search results in the sequencing results as the target search results, wherein N is an integer greater than or equal to 1.

9. The method of claim 7, determining the target search result from the ranking result, comprising:

and determining the target search result according to the at least one reference search result and the sorting result.

10. The method of claim 9, determining the target search result from the at least one reference search result and the ranking result, comprising:

and determining the M search results and the P reference search results as the target search result.

11. A search apparatus, comprising:

the first acquisition module is used for acquiring search words;

12. The apparatus of claim 11, wherein, applied to a map search engine; the second obtaining module is specifically configured to:

13. The apparatus of claim 12, wherein the second obtaining module is specifically configured to:

14. The apparatus of claim 13, wherein the second obtaining module is specifically configured to:

15. The apparatus of claim 13, wherein the second obtaining module is specifically configured to:

16. The apparatus of claim 12, wherein the second obtaining module is specifically configured to:

17. The apparatus according to any one of claims 11-16, the processing module being specifically configured to:

and determining the target search result according to the sorting result.

18. The apparatus of claim 17, the processing module being specifically configured to:

19. The apparatus of claim 17, the processing module being specifically configured to:

20. The apparatus of claim 19, the processing module being specifically configured to:

21. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-10.

22. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1-10.