The message advises method and system
Technical field
The present invention relates to information retrieval field, particularly a kind of message advises method and system.
Background technology
Query suggestion, be called Query Suggestion again, Fig. 1 is the interface synoptic diagram of existing query suggestion system (search engine), as shown in Figure 1, when the user imports the term of " also pearl " at the input area of search engine, search engine can be automatically provides a plurality of query words 11 as " also pearl sound of laughing " with the form of drop-down table, " also pearl sound of laughing this swallow flies lightly " etc., to improve the efficient of user search information, the query suggestion system has become and has reduced the important way that the user obtains information costs at present, for example 30% of Baidu search flow derives from suggesting system for wearing, whether well-done the query suggestion system is, will directly have influence on the user and experience.
Yet, traditional query suggestion system all unites and is shown as the master with text, it exists information limited, friendly inadequately alternately, and the user also needs just can find by search page the problem of the page of wanting, accordingly, the way of some innovations has appearred in industry, for example with entity storehouse (TV library information, movie library information) combine with the query suggestion system, Fig. 2 is the existing query suggestion system interface figure that combines the entity library information, as shown in Figure 2, if the query word in the query suggestion system mates in the entity storehouse to some extent, then can show corresponding information 12 on the right side of query suggestion system interface, though, this suggesting system for wearing that combines the entity library information can have been showed more information, make alternately more horn of plenty, but its shortcoming is only under the situation that the standard queries word of the term of user input and query suggestion systemic presupposition mates fully, corresponding information 12 just can display in the entity storehouse, if user's input is and term like the standard queries part of speech, so just can't show corresponding information 12, just exist in this case and recall on the one hand inadequately, also brought puzzled problem to the user on the one hand, in addition, corresponding information 12 in the entity storehouse usually and the standard queries word mate one to one, therefore, the current query suggestion system that combines the entity library information can't show many corresponding information 12 to the user.
Summary of the invention
The object of the present invention is to provide a kind of message advises method and system, query word and object information to user's history excavate, the object information of M high-quality before realizing collecting under the line, avoided the user in retrieving, to produce delay, experience lf being influenced, the object information of preceding M high-quality of each query word directly is provided for the user in retrieving, directly satisfy user's demand, reduce the retrieval cost.
For addressing the above problem, the invention provides a kind of message advises method, comprising:
Each query word that recording user is selected and corresponding object information, and the eigenwert of this object information is set;
Choose the top n object information according to described eigenwert from the object information of each query word correspondence, wherein N is natural number, and N>=2;
Choose preceding M object information according to described eigenwert from the top n object information of each query word, wherein M is natural number, and M<N;
Be a plurality of participles unit with each query word cutting, and set up the corresponding relation of each participle unit and a plurality of query words;
Set up the corresponding relation of preceding M object information of each participle unit and a plurality of query words respectively according to the corresponding relation of each participle unit and a plurality of query words.
Further, in said method, described object information comprises a kind of or combination in any in title, summary, label, broadcast address and the thumbnail address.
Further, in said method, described eigenwert comprises the relevant weights of every object information of reflection and the correlativity of corresponding query word.
Further, in said method, described eigenwert also comprises the clicking rate value of the selection number of times of object information being determined according to the user.
Further, in said method, described eigenwert also comprises according to the user selects the concern rate value definite to the degree of concern of object information.
Further, in said method, described degree of concern comprises that the user is to browsing time and/or the page turning situation of described object information.
Further, in said method, from the object information of each query word correspondence, choose in the top n object information step according to described eigenwert, from the object information of each query word correspondence, choose the forward N of a clicking rate value object information.
Further, in said method, from the top n object information of each query word, choose in preceding M the object information step according to described eigenwert, obtain comprehensive weights according to described clicking rate value, relevant weights and concern rate value, from the top n object information of each query word, choose M forward object information of described comprehensive weights.
Further, in said method, described clicking rate value, relevant weights and concern rate value be weighted to add up by different weights respectively obtain described comprehensive weights.
Further, in said method, described clicking rate value, relevant weights and concern rate value be weighted by 20%, 30% and 50% weights respectively add up.
According to another side of the present invention, a kind of message advises system is provided, comprising:
Logger module is used for each query word and corresponding object information that recording user is selected, and the eigenwert of this object information is set;
The result chooses module, is used for choosing preceding M object information according to described eigenwert from the object information of each query word correspondence, and wherein M is natural number;
Retrieval module, being used for each query word cutting is a plurality of participles unit, and sets up the corresponding relation of preceding M object information of each participle unit and a plurality of query words respectively.
Further, in said system, described result chooses module and comprises:
First unit as a result is used for choosing the top n object information according to described eigenwert from the object information of each query word correspondence, and wherein N is natural number, and N>=2;
Second unit as a result, M object information before being used for choosing from the top n object information of each query word according to described eigenwert, wherein M is natural number, and M<N.
Further, in said system, described retrieval module comprises:
First indexing units, being used for each query word cutting is a plurality of participles unit, and sets up the corresponding relation of each participle unit and a plurality of query words;
Second indexing units is for the corresponding relation of setting up preceding M object information of each participle unit and a plurality of query words according to the corresponding relation of each participle unit and a plurality of query words respectively.
Further, in said system, the object information of described logger module record comprises a kind of or combination in any in title, summary, label, broadcast address and the thumbnail address.
Further, in said system, the clicking rate value that the feature of described logger module setting comprises the relevant weights of every object information of reflection and the correlativity of corresponding query word, determine the selection number of times of object information according to the user or according to user's selection to one or combination in any in the definite concern rate value of the degree of concern of object information.
Further, in said system, described first as a result the unit from the object information of each query word correspondence, choose the forward N of a clicking rate value object information.
Further, in said system, described second as a result the unit obtain comprehensive weights according to described clicking rate value, relevant weights and concern rate value, from the top n object information of each query word, choose M forward object information of described comprehensive weights.
Further, in said system, described second as a result the unit described clicking rate value, relevant weights and concern rate value be weighted to add up by different weights respectively obtain described comprehensive weights.
Further, in said system, described second as a result the unit described clicking rate value, relevant weights and concern rate value be weighted by 20%, 30% and 50% weights respectively add up.
Compared with prior art, each query word and corresponding object information that the present invention selects by recording user, and the eigenwert of this object information is set, and query word and the object information of user's history excavated, can guarantee that the object information that finally obtains more can meet consumers' demand.
In addition, the present invention chooses the top n object information according to described eigenwert from the object information of each query word correspondence, wherein N is natural number, and N>=2, M object information before choosing from the top n object information of each query word according to described eigenwert again, wherein M is natural number, and M<N, realize in retrieving, directly providing for the user object information of preceding M high-quality of each query word, directly satisfied user's demand, reduced the retrieval cost.
In addition, the present invention is by being a plurality of participles unit with each query word cutting, and set up the corresponding relation of each participle unit and a plurality of query words, set up the corresponding relation of preceding M object information of each participle unit and a plurality of query words at last respectively according to the corresponding relation of each participle unit and a plurality of query words, the object information of M high-quality before can realizing collecting under the line.
Description of drawings
Fig. 1 is the interface synoptic diagram of existing query suggestion system;
Fig. 2 is the existing query suggestion system interface figure that combines the entity library information;
Fig. 3 is the process flow diagram of the message advises method of one embodiment of the invention;
Fig. 4 is the high-level schematic functional block diagram of the message advises system of one embodiment of the invention.
Embodiment
For above-mentioned purpose of the present invention, feature and advantage can be become apparent more, the present invention is further detailed explanation below in conjunction with the drawings and specific embodiments.
As shown in Figure 1, the invention provides a kind of message advises method, comprising:
Step S1, each query word that recording user is selected and corresponding object information, and the eigenwert of this object information is set, here excavate by query word and object information to user's history, can guarantee that the object information that finally obtains more can meet consumers' demand, concrete, the each time behavior of user on search engine, a query requests is sent in the capital to the backstage, query time that can the recording user request in this step, and the displaying situation of each user search is the query word of record request, and with the request responding content namely corresponding object information detail record get off, query word can corresponding a plurality of object informations, an object information also might be followed a plurality of query word correspondences, described object information comprises title, summary, label (tag), a kind of or combination in any among broadcast address (playing url) and thumbnail address (the thumbnail url), the described eigenwert that arranges can comprise the relevant weights of every object information of reflection and the correlativity of corresponding query word, the concern rate value that the clicking rate value of the selection number of times of object information being determined according to the user and select according to the user is determined the degree of concern of object information, wherein, described degree of concern can comprise that the user is to back search behaviors such as browsing time of described object information and/or page turning situations, if the time that the user browses certain object information is more long, page turning is more many, and the concern rate value that this object information then can be set is more high;
Step S2, from the object information of each query word correspondence, choose the top n object information according to described eigenwert, wherein N is natural number, and N>=2, concrete, can in the object information of each query word correspondence, choose the forward N of a clicking rate value object information, for example can in the object information of each query word correspondence, choose 10 forward object informations of clicking rate value, certainly, also can pass through the eigenwert that other sortord obtains, and at N forward object information of eigenwert described in the object information of each query word correspondence;
Step S3, from the top n object information of each query word, choose preceding M object information according to described eigenwert, wherein M is natural number, and M<N, thereby in retrieving, can directly provide the object information of preceding M high-quality of each query word for the user, directly satisfy user's demand, reduce the retrieval cost, concrete, can be according to described clicking rate value, relevant weights and concern rate value are obtained comprehensive weights, from the top n object information of each query word, choose M forward object information of described comprehensive weights, for example can from preceding 10 object informations of each query word, choose 3 forward object informations of described comprehensive weights, wherein, can be with described clicking rate value, relevant weights and concern rate value are weighted to add up by different weights respectively obtains described comprehensive weights, for example with described clicking rate value, relevant weights and concern rate value are respectively by 20%, 30% and 50% weights are weighted and add up.
Step S4, be a plurality of participles unit with each query word cutting, and set up the corresponding relation of each participle unit and a plurality of query words, concrete, can be a plurality of littler participle unit with the query word cutting, comprise along the littler participle unit in the query word as prefix, infix, suffix or other, set up the corresponding relation of a participle unit and a plurality of query words then, for example set up the inverted index of each prefix, infix, suffix or littler participle unit to 10 query word;
Step S5, set up the corresponding relation of preceding M object information of each participle unit and a plurality of query words respectively according to the corresponding relation of each participle unit and a plurality of query words, concrete, behind the corresponding relation of having set up participle unit-a plurality of query words-preceding M object information, can be according to the participle unit of user's input, return the inventory of a query word and object information to the user, 10 query words for example can have been comprised in the inventory, and preceding 3 object informations of each query word correspondence, like this for each query word, by from the object information of query word, taking out earlier preceding 10 object informations, and then in these 10 object informations, find 3 object informations of most possibly meeting consumers' demand, these 3 object informations namely can be used as the suggestion exhibition information of user's inquiry, and each object information can comprise title, summary, label (tag), a kind of or combination in any among broadcast address (playing url) and thumbnail address (the thumbnail url).
The object information of M high-quality had avoided the user to produce delay, experience lf being influenced in retrieving before step S4 and step S5 can realize collecting under the line.
The present invention proposes a kind of message advises method based on user's historical query information excavating, query word and corresponding historical retrieving informations such as object information that this method is selected by analysis user, N bar object information before from the object information of each query word, extracting, and by analyzing the eigenwert based on the object information of user behavior, the object information of M high-quality before from the top n object information of each query word, choosing, and the next door of each query word such as right side show should before M object information, each object information can comprise title, summary, label (tag), a kind of or combination in any among broadcast address (playing url) and thumbnail address (the thumbnail url) can directly be clicked object information when having realized user's browse queries word.
As shown in Figure 4, the present invention also provides another kind of message advises system, comprises that logger module 1, result choose module 2 and retrieval module 3.
Logger module 1 is used for each query word and the corresponding object information that recording user is selected, and the eigenwert of this object information is set, excavate by query word and object information to user's history, can guarantee that the object information that finally obtains more can meet consumers' demand, concrete, the object information of described logger module 1 record comprises title, summary, label, a kind of or combination in any in broadcast address and the thumbnail address, the feature that described logger module 1 arranges comprise the relevant weights of every object information of reflection and the correlativity of corresponding query word, one or combination in any in the clicking rate value of the selection number of times of object information being determined according to the user or the concern rate value of selecting the degree of concern of object information is determined according to the user.
M object information before the result chooses module 2 and is used for choosing from the object information of each query word correspondence according to described eigenwert, wherein M is natural number, and is concrete, described result chooses module 2 and comprises first unit 22 as a result, unit 21 and second as a result.
First as a result unit 21 be used for choosing the top n object information according to described eigenwert from the object information of each query word correspondence, wherein N is natural number, and N>=2, thereby in retrieving, can directly provide the object information of preceding M high-quality of each query word for the user, directly satisfy user's demand, reduce the retrieval cost, concrete, described first as a result unit 21 can from the object information of each query word correspondence, choose the forward N of a clicking rate value object information.
Second M object information before unit 22 is used for choosing from the top n object information of each query word according to described eigenwert as a result wherein M is natural number, and M<N, concrete, described second unit 22 can be according to described clicking rate value as a result, relevant weights and concern rate value are obtained comprehensive weights, from the top n object information of each query word, choose the object information of M forward high-quality of described comprehensive weights, described second as a result unit 22 with described clicking rate value, relevant weights and concern rate value are weighted to add up by different weights respectively obtains described comprehensive weights, for example with described clicking rate value, relevant weights and concern rate value are respectively by 20%, 30% and 50% weights are weighted and add up.
It is a plurality of participles unit that retrieval module 3 is used for each query word cutting, and set up the corresponding relation of preceding M object information of each participle unit and a plurality of query words respectively, the object information of M high-quality before can realizing like this collecting under the line, avoid the user in retrieving, to produce delay, experience lf being influenced, wherein, described retrieval module 3 comprises first indexing units 31 and second indexing units 32.
It is a plurality of participles unit that first indexing units 31 is used for each query word cutting, and sets up the corresponding relation of each participle unit and a plurality of query words.
Second indexing units 32 is used for setting up respectively according to the corresponding relation of each participle unit and a plurality of query words the corresponding relation of preceding M object information of each participle unit and a plurality of query words.
Each query word and corresponding object information that the present invention selects by recording user, and the eigenwert of this object information is set excavate query word and the object information of user's history, can guarantee that the object information that finally obtains more can meet consumers' demand.
In addition, the present invention chooses the top n object information according to described eigenwert from the object information of each query word correspondence, wherein N is natural number, and N>=2, M object information before choosing from the top n object information of each query word according to described eigenwert again, wherein M is natural number, and M<N, realize in retrieving, directly providing for the user object information of preceding M high-quality of each query word, directly satisfied user's demand, reduced the retrieval cost.
In addition, the present invention is by being a plurality of participles unit with each query word cutting, and set up the corresponding relation of each participle unit and a plurality of query words, set up at last the corresponding relation of object information of preceding M high-quality of each participle unit and a plurality of query words respectively according to the corresponding relation of each participle unit and a plurality of query words, M object information before can realizing collecting under the line avoided the user to produce delay, experience lf being influenced in retrieving.
Each embodiment adopts the mode of going forward one by one to describe in this instructions, and what each embodiment stressed is and the difference of other embodiment that identical similar part is mutually referring to getting final product between each embodiment.For the disclosed system of embodiment, because corresponding with the embodiment disclosed method, so description is fairly simple, relevant part partly illustrates referring to method and gets final product.
The professional can also further recognize, unit and the algorithm steps of each example of describing in conjunction with embodiment disclosed herein, can realize with electronic hardware, computer software or the combination of the two, for the interchangeability of hardware and software clearly is described, composition and the step of each example described in general manner according to function in the above description.These functions still are that software mode is carried out with hardware actually, depend on application-specific and the design constraint of technical scheme.The professional and technical personnel can specifically should be used for using distinct methods to realize described function to each, but this realization should not thought and exceeds scope of the present invention.
Obviously, those skilled in the art can carry out various changes and modification to invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these revise and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these change and modification.