CN105808590B - Search engine implementation method, searching method and device - Google Patents

Abstract

The present invention discloses a kind of search engine implementation method, searching method and device.This method comprises: obtaining the objective result of the search expression history and selection that input in user one continuous search process；Expression keyword sequence is determined according to search expression history；It include the connection relationship and weight expressed between keyword and between expression keyword and objective result in semantic net according to expression keyword sequence and objective result generative semantics net.Search engine implementation method, searching method and the device that the disclosure provides, the objective result of expression keyword and selection based on user's input carries out statistics and establishes semantic net, it obtains between keyword and keyword and the distance between keyword and objective result and weight relationship, when user searches for objective result, service can be provided for user based on the semantic net, user's search service information time is saved, provides more acurrate efficient information service for user.

Description

Search engine implementation method, searching method and device

Technical field

The present invention relates to internet area more particularly to a kind of search engine implementation methods, searching method and device.

Background technique

Search engine technique is as Internet technology development gets up, and main function is the feelings in information explosion Under condition, the retrieval of massive information is coped with, meets the needs of people obtain information.

It is to calculate user using all kinds of advanced algorithms to input keyword or ask in terms of one core of search engine technique The correlation with information in backstage mass data index is inscribed, the algorithm taken can evaluate numerous information and key from all angles The relationship of word, evaluation result are exactly list of the information according to sizes related.After user obtains this list, oneself is needed Actively screen suitable search result.User gradually understands the application method of search engine during filtered list, understands The skill of keyword input, the input that can adjust keyword carry out expressing information demand again, take part in from people for this angle The characteristics of search process, the information service systems such as search engine, service ability are sightless for user.But the prior art Need to carry out the judgement of result by user the application method of learning system and the keyword of adjustment input, to reach search mesh , search data means are not relatively intelligent, and user's acquisition time is long, search experience is low, more wasteful user time.

Therefore, it is necessary to propose a kind of search engine implementation method, user can be saved and search for information time, provided for user More acurrate efficient information service.

Summary of the invention

The disclosure technical problem to be solved is how to provide a kind of search engine to realize to save user and search for letter The time is ceased, provides more acurrate efficient information service for user.

The disclosure provides a kind of search engine implementation method, comprising:

Obtain the objective result of the search expression history and selection that input in user one continuous search process；According to searching Rope expression formula history determines expression keyword sequence；According to expression keyword sequence and objective result generative semantics net for searching Index is held up, and includes the connection relationship and power expressed between keyword and between expression keyword and objective result in semantic net Weight.

Optionally, the target knot of the search expression history and selection that input in user one continuous search process is obtained Fruit includes: to obtain the expression keyword of user's input, the expression keyword of adjustment and the target knot selected according to search result Fruit.

Optionally, before according to expression keyword sequence and objective result generative semantics net further include: determine service field, Service field includes sole user, group, vertical field or integrated information service field；Judge that user inputs under service field Expression keyword sequence in express keyword and according to expression Keyword Selection objective result process whether be completely to believe Cease service process.

It optionally, include that the connection between expression keyword and expressed between keyword and objective result is closed in semantic net System and weight, comprising:

If the expression keyword of user's input is unsatisfactory for semantic web updates rule, cast out the expression keyword；

If the expression keyword of user's input meets semantic web updates rule, the expression keyword and semantic net of user's input The similarity of middle keyword is less than similarity threshold, in the expression keyword that semantic net addition user inputs as in semantic net New keywords, and the objective result of user's selection is recorded, generate or update other keywords in the new keywords and semantic net Between connection relationship weight and the new keywords and objective result multi-stress；

If the expression keyword of user's input meets semantic web updates rule, the expression keyword and semantic net of user's input The similarity of middle keyword be more than or equal to similarity threshold, generate or update semantics net in the synthesis of keyword and objective result because Son, multi-stress are determined according to the shortest path factor and weight factor of the keyword in semantic net to objective result, Weight factor is that the number normalization determination of objective result is reached according to keyword in the semantic net of historical statistics.

Optionally, judge to express keyword in the expression keyword sequence of user's input under service field and be closed according to expression Whether the process of keyword selection target result is complete information service process, comprising: determines that user inputs expression keyword and searches Number of clicks of the process of rope objective result to each objective result；If the number of clicks of objective result is greater than information service threshold value, Service process is then determined as complete service process；

And/or

The process that user inputs the objective result selected after expression keyword is determined as complete service process.

The disclosure additionally provides a kind of searching method, comprising: determines the expression keyword that user inputs in expression formula；It determines Express keyword matched start node in semantic net, wherein semantic net is based on the expression keyword in user's search history It is generated with objective result；Objective result is determined according to start node in semantic net, objective result is according to start node to target As a result distance and weight determines；Prompt objective result.

Optionally, this method further include: prompt keyword is determined according to start node in semantic net；Prompt prompt is crucial Word.

Optionally, determine that prompt keyword includes: according to start node in semantic net

If start node is not the endpoint node of semantic net, selects start node to develop to the path of objective result, determine Node on path is as prompt keyword.

Optionally, if start node is the endpoint node of semantic net, the corresponding objective result of start node is prompted to use Family；If start node is not the endpoint node of semantic net, the weight factor of selection target result is greater than the target knot of weight threshold Fruit；It selects start node to develop to the path factor of objective result and is greater than the path of path threshold；Determine other sections on path Point is as prompt keyword；Prompt prompt keyword and/or multiple objective results；Wherein, the node in semantic net be it is identical or Similar crucial phrase at set, weight factor be according in the semantic net of historical statistics keyword reach objective result time Number normalization determination.

A kind of search engine realization device of the disclosure, comprising: module is obtained, for obtaining one continuous search process of user The search expression history of middle input and the objective result of selection；Determining module, for determining table according to search expression history Up to keyword sequence；Generation module draws for being used to search for according to expression keyword sequence and objective result generative semantics net It holds up, includes the connection relationship and weight expressed between keyword and between expression keyword and objective result in semantic net.

Optionally, obtain module be used for obtain user input expression keyword, adjustment expression keyword and according to The objective result of search result selection.

Optionally, the device further include:

Service field determining module, for determining that service field, service field include sole user, group, enterprise or row Industry；Information service process judgment module, it is crucial for judging to express in the expression keyword sequence of user's input under service field Word and according to expression Keyword Selection objective result process whether be complete information service process.

Optionally, evaluation module is cast out if the expression keyword for user's input is unsatisfactory for semantic web updates rule The expression keyword；

If the expression keyword that evaluation module is also used to user's input meets semantic web updates rule, the expression of user's input The similarity of keyword is less than similarity threshold in keyword and semantic net, in the expression keyword of semantic net addition user's input As the new keywords in semantic net, and the objective result of user's selection is recorded, generates or update the new keywords and semanteme The multi-stress of connection relationship weight and the new keywords and objective result in net between other keywords；

If the expression keyword that evaluation module is also used to user's input meets semantic web updates rule, the expression of user's input The similarity of keyword is more than or equal to similarity threshold, keyword and mesh in generation or update semantics net in keyword and semantic net The multi-stress of result is marked, multi-stress is the shortest path factor and power according to the keyword in semantic net to objective result What repeated factor determined, weight factor is to be normalized really according to the number of keyword arrival objective result in the semantic net of historical statistics Fixed.

Optionally, information service process judgment module is used for: determining that user inputs expression key word search targets result Number of clicks of the process to each objective result；If the number of clicks of objective result is greater than information service threshold value, by service process It is determined as complete service process；And/or the process that user inputs the objective result selected after expression keyword is determined as Whole service process.

A kind of searcher of the disclosure, comprising: expression key word analysis module, for determining that user inputs in expression formula Express keyword；Start node analysis module, for determining expression keyword matched start node in semantic net, wherein Semantic net based in user's search history expression keyword and objective result generate；Determining module, for the root in semantic net Determine that objective result, objective result are determined according to the distance and weight of start node to objective result according to start node；Prompt mould Block, for prompting objective result.

Optionally, determining module is also used in semantic net determine prompt keyword according to start node；Cue module is also For prompting keyword.

Optionally, cue module is used for: if start node is not the endpoint node of semantic net, start node evolution being selected to arrive The path of objective result determines the node on path as prompt keyword.

Optionally, if cue module is used for the endpoint node that start node is semantic net, by the corresponding target of start node As a result it is prompted to user；If and/or cue module for start node be not semantic net endpoint node, selection target result Weight factor is greater than the objective result of weight threshold；It selects start node to develop and is greater than path threshold to the path factor of objective result The path of value；Determine other nodes on path as prompt keyword；Prompt prompt keyword and/or multiple objective results, Wherein, the node in semantic net be the same or similar crucial phrase at set, weight factor is according to the language of historical statistics Keyword reaches the number normalization determination of objective result in justice net.

Search engine implementation method, searching method and the device that the disclosure provides, the expression based on user's input are crucial Word and the objective result of selection carry out statistics and establish semantic net, obtain between keyword and keyword and keyword and target knot The distance between fruit and weight relationship can provide service letter based on the semantic net when user searches for objective result for user Breath saves user and searches for information time, provides more acurrate efficient information service for user.

Detailed description of the invention

Fig. 1 shows the flow chart of the search engine implementation method of one embodiment of the invention.

Fig. 2 shows the schematic diagrames that the continuous search process of one embodiment of the invention obtains objective result.

Fig. 3 shows the flow chart of the search engine implementation method of another embodiment of the present invention.

The search engine that Fig. 4 shows one embodiment of the invention realizes the flow chart of optimization method.

Fig. 5 shows the flow chart of the searching method of one embodiment of the invention.

Fig. 6 shows the flow chart of the searching method of another embodiment of the present invention.

Fig. 7 shows the flow chart of the searching method of another embodiment of the invention.

Fig. 8 shows the schematic diagram of the semantic net of a sphere of learning one professor group of one embodiment of the invention.

Fig. 9 shows the structural schematic diagram of the search engine system of one embodiment of the invention.

Figure 10 shows the structural block diagram of the search engine realization device of one embodiment of the invention.

Figure 11 shows the structural block diagram of the search engine realization device of another embodiment of the present invention.

Figure 12 shows the structural block diagram of the searcher of one embodiment of the invention.

Figure 13 shows the structural block diagram of the search engine realization device of another embodiment of the invention.And

Figure 14 shows the structural block diagram of the searcher of another embodiment of the invention.

Specific embodiment

With reference to the accompanying drawings to invention is more fully described, wherein illustrating exemplary embodiment of the present invention.

Search engine is for the core objective that the understanding of customer problem is that search engine is pursued.Under normal circumstances, it is searching for In the service process of engine, there are three core links.First core link is the understanding for problem, and search engine passes through each Class algorithm and method understand that user requests to put question to the core point of (QUERY), it is known that user looked for is what.Second core link Be: how tissue being carried out to background information resource, be indexed, make background information by it is a kind of can expressing information essence in a manner of and The mode for adapting to information inquiry is stored；Third core link is: search need and search result directly match, and make user It is matched between demand and storage information, and carries out reasonable computation, calculated result is to face the demand of user information, and reflect The information of Essence of Information expression.

Basic thought of the embodiment of the present invention is to introduce semantic network technology in information service fields such as search engines, proposes one Kind can for individual, group, field application search engine realize method, this method formation be that one kind can perceive field Variation and a certain section of period class problem expression variation semantic network, and the semantic net can according to individual, group, The application autonomous evolution of the information service in field, ultimately form towards particular person, group, field understanding service-Engine.According to this Semantic network can overcome different user to be directed to the otherness of same thing difference description, even if different in user's search process User uses different expression for same thing, it is also possible to obtain identical objective result or prompt keyword.The present invention The method of embodiment be also equipped with according to time and information requirement, expression, generation variation, understand the ability of information, adapt to different The search in period is expressed and search understands and objective result recommends ability.

Fig. 1 shows the flow chart of the search engine implementation method of one embodiment of the invention.As shown in Figure 1, this method master Include:

Step S100 obtains the target knot of the search expression history and selection that input in user one continuous search process Fruit；

In one embodiment, the expression keyword of user's input, the table of adjustment can according to the search process of user, be obtained Up to keyword and the objective result selected according to search result.

For same problem, the identification of different expression ways is recorded using the search history of user, Learning Service success institute Corresponding history input expression keyword, records the expression keyword continuously inputted and corresponding target knot if servicing successfully Fruit.Search engine system searches for the user in certain time the record of input, and user searches for input expression in analysis record Adjustment, the final goal result and this corresponding mesh that study user selects in several continuous input processes whithin a period of time Mark the different expression keywords of result.

It should be noted that search expression of the invention not merely refers to input text, those skilled in the art can be with Understand, user can input search expression in many ways, for example, it may be the modes such as audiovideo.For example, being based on language In the search process of sound dialogue, for a problem, there are many Expression of language for the possible energy of user, if user is according to system Search result has selected objective result, can generally stop search and check objective result, user voice can be expressed and be selected Objective result gets off as log recording, to statistically analyze generative semantics net.

Step S102 determines expression keyword sequence according to search expression history；

In search routine, user need to learn the knowledge in strange field on one side, while continuous adjustment input expression, can be with Information required for being obtained in a strange field, the target knot of record and keyword and selection in study user's search process Fruit has different expression to same problem by learning different user, records the process of multiple user's search, can extract pass Connection relationship between keyword and keyword.

Step S104, according to expression keyword sequence and objective result generative semantics net to be used for search engine, semantic net In include expression keyword between and expression keyword and objective result between connection relationship and weight.

In one embodiment, it at a distance from objective result and can be expressed crucial and objective result according to expression keyword Weight determines the multi-stress of the keyword.

The multi-stress that expression keyword path and keyword and objective result weight determine decides a start node How to obtain reaching the target information finally provided for user.On the one hand, the connection relationship between analysis of key word and keyword, The shortest path that a keyword is connected to another keyword is obtained, on the other hand, calculates keyword linking objective result at certain Weighted value on path, path weight value is established during semantic net and after foundation in use process, according to new user's table It reaches and the selection of objective result is continued to optimize obtained by adjustment.

In one embodiment, several users be can recorde in search process several times, record user according to a table It is adjusted to another expression keyword number up to keyword, and using the number as connection weight.The connection weight refers to from one The number that a keyword is selected to next keyword, user, the weight obtained by being weighted or normalizing.

In one embodiment, which is according to path and weight between the keyword and objective result in semantic net Composite weighted determines.The relationship of expression keyword and objective result can be determined according to the multi-stress in this way.In search process In, if user's input is the keyword, service is provided for user according to the multi-stress.

In one embodiment, by learning multiple search process of multiple users, input keyword based on multiple users, The objective result for adjusting keyword and selection, can be generated the relational network of keyword sequence and objective result, can should Network is referred to as semantic net.When generating the semantic net, the connection relationship of each expression keyword is recorded and according to expression key Word obtain objective result number, and through historical statistics by the number normalize as the weight of the keyword and objective result because Son.

The semantic network of the embodiment of the present invention is according to the network of expression keyword and objective result composition, it is according to key The meaning of word forms one " keyword and objective result network ", and each node in the keyword network all with adjacent segments Point has connection relationship, and keyword is the basic component of the semantic net, which follows lexical semantics principle, can be with Carry out expression node using synonym collection, and these keyword nodes are associated with a certain number of relationship types, constitutes One keyword and objective result semantic net.The semantic net can be can be applied to natural language by the semantic net formed In understanding field, machine translation, input method field and searching engine field.

The search engine implementation method of the embodiment of the present invention, the target knot of expression keyword and selection based on user's input Fruit carries out statistics and establishes semantic net, obtains between keyword and keyword and the distance between keyword and objective result and weighs Series of fortified passes system provides service based on the semantic net in user's search process for user, can save user and search for information time, More acurrate efficient information service is provided for user.

In one embodiment, for user before obtaining final result, the information of input may be variation.User is at one In continuous information access process, the result returned there are one according to system learns the process how to express；User passes through Solution returns the result, and also has the process of cognition to oneself demand.By record user's input, input correspondence is returned the result every time And final result establishes a semantic network or semantic network database after carrying out feature extraction.

Fig. 2 shows the schematic diagrames that the continuous search process of one embodiment of the invention obtains objective result, as shown in Fig. 2, In input 1, into input n, user inputs different expression keywords and wishes to search the good market for buying general merchandise, inputs knows for the first time Name general merchandise shows multiple objective results, and user and non-selected any search show that user adjusts as a result, in second of input Expression inputs other several keywords, and multiple search results of display, use is non-selected per family, in n-th input, user's input Famous general merchandise has Yaohan as the result is shown as a result, user clicks the first Yaohan of Shanghai, into related store information in search The page obtains the information that user needs.General merchandise, famous general merchandise, Yaohan and the first Yaohan of Shanghai form a search history Record, wherein general merchandise, famous general merchandise, Yaohan belong to the expression keyword continuously inputted, and the first Yaohan of Shanghai belongs to target As a result.

Referring to Fig. 2, the expression keyword that general merchandise, famous general merchandise, Yaohan continuously input as user is determined as key Word sequence.

Based on multiple statistical result, weight of the general merchandise to famous general merchandise is obtained, the weight of famous general merchandise to Yaohan, with And Yaohan is to the weight of the first Yaohan of Shanghai.

Based on the semantic net that Fig. 2 is formed, when a user uses the search engine or input method based on the semantic net, if It searches for " general merchandise ", search engine can prompt keyword " famous general merchandise ", " Yaohan " according to the semantic net, when the user clicks really After fixed search " general merchandise ", browser preferentially exports " the first Yaohan of Shanghai " relevant search information, for users to use.

The search engine implementation method of the embodiment of the present invention can solve people for the difference in problem understanding, description Property, the adaptability that individual difference brings standard product is insufficient.It is excavated, is understood based on the record in search engine interactive process Service field knowledge and concept solve the problems, such as to need multi-party adjustment that could obtain objective result when user search request.

Fig. 3 shows the flow chart of the search engine implementation method of another embodiment of the present invention, as shown in figure 3, this method Include:

Step 300, the target knot of the search expression history and selection that input in user one continuous search process is obtained Fruit；

Step 301, expression formula keyword is determined according to search expression history；

Step S302 determines that service field, service field include sole user, group etc.；

For example, in information service field, individual, group, vertical field or a comprehensive service can be directed to Field, information service system record user and input keyword and final choice result.This service field can be traditional interconnection Net field, mobile Internet field are also possible to voice information services platform.

The objective result information that user inputs keyword and corresponding selection is recorded, in this way by the accumulation of long a period of time, Keyword can be obtained and broadcast a corresponding relationship net of enterprise, search result information or objective result.Further, may be used To record the variation that user inputs keyword during user's use information service system.

Step S303 judges to express keyword under service field in the expression keyword sequence of user's input and according to expression Whether the process of Keyword Selection objective result is complete information service process.

In one embodiment, judge to express in expression keyword sequence that user under service field inputs keyword and according to Whether the process for expressing Keyword Selection objective result is complete information service process, comprising: within a time cycle, such as 1 day, 1 month, 1 year etc., determine that user inputs click time of the process to each objective result of expression key word search targets result Number；If the number of clicks of objective result is greater than information service threshold value, service process is determined as complete service process.

In one embodiment, the process that user can be inputted to the objective result selected after expression keyword is considered as completely Service process.Can judge whether service process is complete service process by other methods, as user clearly expresses at present Search result is exactly the objective result that it wants to obtain；Or user stopped search after selecting some objective result.

One complete information service process can be expressed as follows, and user inputs keyword first, then observation search knot Fruit list, if that user clicks in lists as a result, each search result and click search result that record user clicks Number N, N >=0.In the present invention, it is the information service factor that we, which can define N, during generic services, when N is not zero, It is thought that a complete information service process；Under certain specific environments or in a certain field, N can take one of system Empirical value is considered a complete information service process greater than this empirical value, less than this empirical value not think be One complete information service process.During a complete information service, user converts keyword, but does not generate There are many reason of search result.

For example, by voice information services platform attend a banquet attendant search process for, it is preceding in customer service system Refer to user and makes a phone call or by the users of other access means inquiry messages, the user that attends a banquet refers to contact staff, according to user With user exchange and the understanding of oneself, input keyword and some feature representations, scan for letter in background service information Breath.This transformation keyword may be generated by following reason:

1, forward direction user for information representation otherness, the expression of forward direction user be its according to itself to required problem The understanding of description, and user's (attending a banquet) understands whether expression is correct since may and not know about this in the service, if Backstage stores corresponding information, therefore causes keyword search for the first time and do not find satisfactory result.

2, user's (attending a banquet) by the understanding to system and interacts the product for having adjusted input keyword, and relying on service experience It is tired, the core point of (forward direction user) information requirement is taken out, but there are certain variations in service process for this central point.

In this way by long-term accumulation, the corresponding relationship that keyword and objective result for example broadcast enterprise can be obtained Net.The last layer of this net is the keyword that end user's (attending a banquet) finally enters, and the corresponding user of this keyword (sits Seat) selection final goal information.By long-term study and optimization, based on multiple record and adjustment, when this Information Network After network reaches a certain level, so that it may be applied in reality system.No result may be broadcasted when user inputs those Keyword after, for example, according to conventional information search or search method, may without casting result keyword after, according to language Justice net obtains keyword casting corresponding with this keyword enterprise of finish node.

Step S304 includes that expression is crucial according to expression keyword sequence and objective result generative semantics net, in semantic net Connection relationship and weight between word and between expression keyword and objective result.

It in one embodiment, include between expression keyword and between expression keyword and objective result in semantic net Connection relationship and weight, comprising:

If the expression keyword of user's input is unsatisfactory for semantic web updates rule, cast out the expression keyword；If with The expression keyword of family input meets semantic web updates rule, the phase of the expression keyword and keyword in semantic net of user's input It is less than similarity threshold like degree, in the expression keyword that semantic net addition user inputs as the new keywords in semantic net, and The objective result for recording user's selection, the connection for generating or updating in the new keywords and semantic net between other keywords are closed It is the multi-stress of weight and the new keywords and objective result；If the expression keyword of user's input meets semantic net more New rule, the similarity of keyword is more than or equal to similarity threshold in the expression keyword and semantic net of user's input, generate or The multi-stress of keyword and objective result in update semantics net, multi-stress are according to the keyword in semantic net to target knot What the shortest path factor and weight factor of fruit determined, objective result is reached according to keyword in semantic net through historical statistics Number, which normalizes, determines weight factor.

In one embodiment, weight factor is the number for reaching objective result according to keyword in semantic net, by history Statistics, normalization are calculated and are obtained.It is also conceivable to other factors determine the weight factor, such as time factor is considered, at one Period often as a result, might not be more in next period.

In one embodiment, semantic web updates rule can be newly-increased keyword rule, retain keyword rule, delete and close Keyword rule, more new keywords multi-stress rule etc..

In one embodiment, if the keyword that user newly inputs, it is unsatisfactory for semantic web updates rule, then system casts out this Keyword；Newly-increased node rule can be described below: 1, whether user obtains objective result in a series of continuous inputs；2, The correlation judgement of continuous input content；Or dispersion judgement；Whether continuous input is directed to the same goal description, or continuous Input it is conceptually whether identical, or continuous input in expression whether similar equal 3, user input whether key can be extracted The expression etc. that word or system can reuse；If user's input meets the technology judgement that semantic net increases node newly；Meanwhile it using Original crucial Word similarity is less than similarity threshold in family input expression keyword and semantic net, it can judges that this is one New user's expression, such as: keyword then adds the expression such as keyword of user's input as in semantic net in semantic net New keywords, and record the objective result of user's selection.

In one embodiment, letter is set for individual, group, vertical field, area or a selected service orientation Cease service factor.And according to this information service factor, selected information service process.

During recording an information service, what the keyword and correspondence of user's input selected in search result list is searched Hitch fruit.

In one embodiment, semantic web updates rule can be screened by keyword and be determined, by an evaluation function, be sentenced Break this word whether can be kept as semantic net increase newly node.

In one embodiment, this evaluation function can take following principle:

1, in the service dictionary of service accumulation, judge that user inputs whether expression keyword is a complete word, or It whether is a complete expression.Judge there are many algorithms, method is segmented for example, by using basic binary or ternary, after judging participle Word and dictionary in word it is whether identical.Here we record this and are evaluated as α₁, wherein 0≤α₁≤ 1, if α₁It is smaller, then illustrate Lower, the α with the Keywords matching in semantic net dictionary₁It is bigger, then illustrate high with the matching degree of keyword in semantic net.

Generally, it is believed that word is to be made of two words or three words are constituted.If it is considered to being made of two words , in short, since lead-in, every two word is divided into a word；Remove lead-in later, as soon as then every two word word, in this way, One group of participle has been obtained, has been divided again after some function words can also be removed in practice.It is in actual use it is also contemplated that continuous between word Property or context relation, and using above-mentioned binary, ternary participle legal principle solution user input expression keyword.

It should be noted that although the present invention illustrates how to understand that user inputs as example using binary, ternary participle method Expression keyword, but the invention is not restricted to this, which is only an exemplary participle illustrating of the present invention Method, in fact, in embodiments of the present invention can be using a variety of segmenting methods or the concept extraction of natural language processing field Or the method for feature extraction, obtain a keyword, or the feature representation of corresponding input.

In one embodiment, if user's input is a complete expression, such as there is the sentence of Subject, Predicate and Object composition, such as " general merchandise of where doing shopping is relatively good " perhaps short sentence such as " famous general merchandise " or short keyword such as " Yaohan ", " famous " " general merchandise ", " famous general merchandise ", " Yaohan " can be then obtained by segmenting method, and then are based on user location " Shanghai " determines that the connection relationship of keyword is " general merchandise " → " famous general merchandise " → " Yaohan ".

In one embodiment, it, based on network address identification and time record, can also can be obtained according to determine semantic net Family location and search time are taken, based on different areas, time, semantic net is adjusted, enables search engine more preferable The relationship for understanding the expression keyword of user and keyword connection relationship, keyword and objective result are established according to keyword.

2, judge whether with the user that has recorded to input expression Keywords matching by user's keyword currently entered that there are one Fixed matching relationship.This keyword recorded can be the keyword of semantic net node in the present invention.It can be using complete Matching and similitude match two kinds of rules, and the selection of specific rules needs to determine according to specific service field.E.g. service It, can be using exact matching rule in the semantic net of personal user；E.g. serve the semantic net of group or vertical field It can be using the rule based on similitude；The bigger semantic net of an e.g. service range, then preferably with similitude rule. Here we record this and are evaluated as α₂, wherein 0≤α₂≤1。

Work as α₁When less than certain threshold value, then it is assumed that this keyword is not the information representation of a standard, and system is cast out This keyword.

Work as α₁Greater than certain threshold value, and α₂When being less than certain threshold value again, it is believed that user's input be one it is qualified can The keyword used by system.Add this node in corresponding position so in semantic net.It records simultaneously subsequent whether related Keyword continues to input, and records the information of subsequent user selection.

Work as α₁Greater than certain threshold value, and α₂When being greater than certain threshold value again, then it is assumed that some of user's input and semantic net Node matches, and at this moment can select answer from this node to end user by calculating acquisition comprehensively considers distance and power The multi-stress of weight, is denoted as β.

It should be noted that α₁、α₂Size be subject to actual selection, different size of value can be selected in different field, The present invention does not do the size of the two values specifically defined.

In one embodiment, multi-stress β is by β₁、β₂What two factors calculating were got.Wherein, factor-beta₁, it is to calculate this Shortest path of a keyword to final casting information.Factor-beta₂, it is the weight for calculating this keyword to final casting information Coefficient, weight here are the number for reaching objective result according to keyword in semantic net, are calculated by historical statistics, normalization And it obtains.It is also conceivable to other factors determine the weight factor, such as time factor is considered, in the knot of a period often Fruit, might not be more in next period.

Optimization algorithm COMPREHENSIVE CALCULATING factor-beta can be used₁And factor-beta₂Multi-stress β.If for example, can select first The dry shortest expression in path selects the maximum expression of path weight value, and then determine multi-stress in the shortest expression in path β；The maximum expression of several weights can also be selected first, and the shortest expression in path is selected in the maximum expression of weight, into And determine multi-stress β；The composite weighted that path and weight can also be comprehensively considered determines multi-stress etc..With multi-stress Height select to arrange two class data for user: the first kind be for user recommendation Search Hints keyword, the second class be for The possible search result that family is recommended.

The degree of correlation basic principle of search engine one keyword of differentiation and webpage is existing using pagerank technology Have and does not account for the calculating process that user used and expressed behavior in technology.It in one embodiment, can be by keyword and target As a result connection relationship and weight relationship is added in the differentiation calculation formula of search service system correlation applied by system, changes The relevant calculation process and result of kind formula.I.e. if occurring this keyword in webpage, it is first considered that its correlation, it is related Degree sorts according to the significance level of webpage, this significance level is calculated by pagerank model, i.e., if a webpage It is in an important website, and more (the also webpage weights of this webpage of consideration link of the quantity for being directed toward or linking this webpage Want degree).

It should be noted that although the present invention is with α₁For threshold value, judge whether the expression of user's input is one normal Expression, uses α₂For threshold value, the similarity of keyword in the keyword and semantic web data library of input is judged.But the present invention couple α₁、α₂Size do not do specific restriction, the size of the two values may be set according to actual conditions in those skilled in the art.

Based on existing search technique, the semantic net that the embodiment of the present invention is formed is an expression keyword and objective result Space net structure, the generation of this reticular structure is that the record by user's use process and refining generates, therefore can be with Improve the calculating process that an objective result list is reached by a keyword, it can improve traditional pagerank technology.

In this way, by the search process of study user, the keyword of keyword, adjustment based on user's input and final The objective result of selection, and establish semantic net by matching degree method and similarity based method includes keyword in the semantic net with The relationship between relationship and keyword and objective result between keyword.

The search engine implementation method of the embodiment of the present invention, available people are for the difference in problem understanding, description Property, the problem that individual difference brings the adaptability deficiency of standard product is eliminated, solution cannot will be formed in practical application Relative normalized semantic net and the problem of with by the semantic net and specific field using combining.

The search engine that Fig. 4 shows one embodiment of the invention realizes the flow chart of optimization method, as shown in figure 4, the party Method includes:

Step S401, service field is determined.

In one embodiment, the first determination field to be serviced, the service field can make some people, a certain group, certain The fields such as one field such as hotel service, medical services, Chinese character input；It can also be directed to a certain pervasive field, do not limited specifically To some specifically field.

Step S402, the information service factor is determined.

Step S403, the system recommendation result for obtaining user's search or using.

Step S404, user's input and characteristic key words similarity calculation.

When executing similarity calculation, step S405 query semantics gateway keyword, information representation library are needed to be implemented, according to language Adopted gateway keyword, information representation library judge the expression and the similarity of keyword in existing semantic net of user's input.

Step S406, judge whether to reach evaluation threshold value.

Whether the expression that user's input can be judged according to semantic net dictionary is a normal expression, can be used similar The judgment method of degree judges whether the keyword of user's input is a normal expression, if it is a normal expression, then Optimize semantic net according to the objective result of the keyword of user's input and selection, if not normal expression, then gives up this Keyword, into common search routine, i.e. execution step S407.

Step S407, it is not up to threshold value, process terminates.

Step S408, optimization semantic net is formed using user's search key and final choice (or non-selected) result.

Step S409, judgement and semantic net have the similarity of keyword.

Step S410, semantic net keyword, information representation library.

Step S411, judge whether to reach semantic net threshold value.

Step S412, strengthen the oriented connection of correlation of semantic net.

Step S413, new node and objective result are added in semantic net.

Different user has different expression to same problem, while service field knowledge is difficult to be easily absorbed on a cognitive level by the user.User needs Learn the knowledge in strange field, while continuous adjustment input expression on one side, it could be required for a strange field obtains Information.Systematic learning searches for user the record of input certain time, and user searches for the variation of input adjustment in analysis record, Learn user in a continuous input process, the different expression of the objective result of the selection of final result and the corresponding selection are closed Keyword.Meanwhile according to the method that this patent embodiment provides or it can also differentiate that backstage can difference between quality of service information Property, according to the user's choice, according to frequency of use, an Optimal scheduling is carried out to information.

For same problem, identify that user's difference expresses keyword, using this historical record, Learning Service success institute is right The history input expression answered such as obtains the history input expression keyword of objective result by the adjustment of keyword, can with backstage Service information content carries out conceptual understanding and feature extraction together, forms a kind of input reminding method based on semantic net or user Possible input keyword prompt and corresponding objective result prompt.

The search engine implementation method of the embodiment of the present invention, based on the conceptual understanding to service field knowledge, search engine The problem of record in interactive process excavates, understands user search request.Understood based on information of the user to input, is used A kind of mode for the semantic net optimizing autonomous evolution optimizes system.

Above-mentioned search engine implementation method can be applied to cell phone client, handheld device, internet device, voice to believe Breath service such as is attended a banquet at search engine services system or the information recommendation system.In this way, by user, groups of users, service field Equal search service systems history serve log is analyzed, and extracting can be valuable used in later user or system Available information.Also it can be applied to Service Source scheduling system etc., can differentiate the type of business of user demand, and according to The characteristics of type of business and current information service reasonably can provide the resource of information service in scheduling backstage, make background service ability It maximizes.In later user service, it can use the above-mentioned valuable available information extracting and later use be provided User experience can be improved in family, enhances the service effectiveness of service enterprise, and then obtain bigger commercial interest, is conducive to simultaneously Service provider carries out more preferable resource distribution, saves cost of serving.The search engine implementation method of the embodiment of the present invention can be according to letter The capability development of the ability, service that cease provider network goes out user's search engine easy to use and related telecommunications information service system System, can develop new product for example similar to products such as Siri, service robots according to the search implementation method.

In one embodiment, propose it is a kind of can for individual, group, field application searching method, this method formed Be a kind of semantic network that can perceive field variation and the variation of a certain section of period class problem expression, according to this net Network can overcome different user to be directed to the otherness of same thing difference description, so that different user is different for same thing Description, it is also possible to obtain identical result.

Fig. 5 shows the flow chart of the searching method of one embodiment of the invention.As shown in figure 5, this method specifically includes that

Step S501, the expression keyword that user inputs in expression formula is determined.

Judge that user inputs whether expression keyword is a complete word, or whether is a complete expression.Judgement There are many algorithms, segments method for example, by using basic binary or ternary, and whether the word in word and dictionary after judging participle is identical.

In one embodiment, the expression of user's input may include multiple keywords, based on the keyword reason in semantic net The meaning for solving the expression of user's input determines the expression keyword that user inputs in expression.

It in one embodiment, can be using such as the evaluation of estimate α in above-described embodiment₂, wherein 0≤α₂≤ 1, if α₂It is smaller, then Illustrate, α lower with the Keywords matching in semantic net dictionary₁It is bigger, then illustrate high with the matching degree of keyword in semantic net.Root The keyword to be searched for of user can be determined according to the evaluation of estimate.

Step S502, expression keyword matched start node in semantic net is determined, wherein the semantic net is based on use Expression keyword and objective result in the search history of family generate.

In one embodiment, semantic net interior joint can be the set of multiple equivalent in meaning or similar keyword.

By understanding the expression of analysis user, after obtaining expression keyword, by the pass in the expression keyword and semantic net Keyword matches, and determines one or more start nodes in semantic net.It can be judged according to above-mentioned similarity based method crucial The matching relationship of keyword in word and semantic net.

The selection of start node has several calculation method.One of calculation method is to calculate the expression inputted and certain Similarity between one node (keyword or certain expression characteristic) just calculates same word between word and word if being all word Number, and whether sequence related, i.e., Keywords matching calculates.If it is expression, can using syntactic analysis, semantic analysis, Comparison between crucial term vector calculates.

In one embodiment, the matching for judging that the expression keyword of user's input whether there is with semantic net interior joint is closed System, semantic net node can be a keyword, be also possible to multiple same or similar crucial combinations.It can use Full matching and similitude matching etc. are regular, and the selection of specific rules needs to determine according to specific service field.

It for example, can be using exact matching rule if it is the semantic net for serving personal user；If it is service Semantic net in group or vertical field can be using the rule based on similitude；If it is the bigger language of a service range Adopted net, then preferably with similitude rule.

It should be noted that in the specific implementation process, can be according to information service resource the case where, flexible choice is similar Property computation rule.

If being greater than certain value according to the similarity of keyword in the expression of user's input determining keyword and semantic net, The keyword in semantic net is selected to provide service as start node, and then for user.

Step S503, objective result is determined according to start node in semantic net, objective result is according to start node to mesh The distance and weight for marking result determine；

Step S504, objective result is prompted.

After start node has been determined in semantic net, according to the relationship in semantic net between keyword and keyword, with And the relationship between keyword and objective result, objective result can be provided for user and be shown on the desplay apparatus, if having more A objective result prompts objective result according to the weight sequencing of objective result.

The searching method of the embodiment of the present invention can analyze the expression formula of user's input, be closed according to the expression of user's input Keyword query semantics net, semantic net are to use according to the connection relationship and keyword of keyword and keyword and the weight of objective result Family prompts keyword and information service, can save user and search for information time, provides more acurrate efficient information clothes for user Business.

Fig. 6 shows the flow chart of the searching method of another embodiment of the present invention.Step S601~step S603 of Fig. 6 with Correspond to that step S501~step S503 is essentially identical, and details are not described herein again for brevity in Fig. 5, as shown in fig. 6, this method is also Include:

Step S604 determines prompt keyword according to start node in semantic net；

Step S605 prompts the user with prompt keyword；

Step S606 prompts objective result.

It has determined in semantic net after start node, has determined that the start node develops to multiple possibility of multiple destination nodes Path provides user's keyword that may think input according to the relationship between keyword and keyword for user, and mention for user Show the keyword.

In one embodiment, if determining that prompt keyword includes: that start node is not according to start node in semantic net The endpoint node of semantic net selects start node to develop to the path of objective result, determines that the node on path is closed as prompt Keyword.

In one embodiment, if start node is the endpoint node of semantic net, the corresponding objective result of start node is mentioned Show to user；If start node is not the endpoint node of semantic net, the weight factor of selection target result is greater than weight threshold Objective result；It selects start node to develop to the path factor of objective result and is greater than the path of path threshold；It determines on path Other nodes are as prompt keyword；Prompt prompt keyword and/or multiple objective results；Wherein, the node in semantic net is The same or similar crucial phrase at set, weight factor be according in the semantic net of historical statistics keyword reach target knot The number normalization determination of fruit.

Objective result can be the Enterprise Lists of service in one embodiment, when user inputs this network arbitrary node After keyword, if this keyword is in the last layer of semantic network, so that it may the enterprise triggered by this keyword List or information on services arrange according to the height of weight or the frequency, this relationship can also be added to applied by system and be searched Rope service system correlation differentiates in calculation formula, improves the relevant calculation process and result of formula.

If the user that attends a banquet, which inputs keyword search, can use businessService information, when this keyword is in the non-most deutomerite of network Point can then calculate and be developed by this node to the path of objective result, obtain one reach on threshold value several have The path of final goal result, while obtaining and having changed keyword on path, at this time based in above-mentioned search engine implementation method Formation semantic net, search engine system can be inputted to user's prompt of attending a banquet with the possible search of this keyword or target As a result.

After inputting keyword by search engine, based on the semantic net that search engine implementation method generates, attends a banquet or use It may be that user provides service or meets the service list of customer information requirement that family, which obtains one, attend a banquet by linking up with user (or user oneself selects in lists) has selected the final company information for user's casting.

In this service process, by the accumulation of serve log, the corresponding casting enterprise of a multipair multi-key word can be obtained The list of industry.Further, due to obtaining the keyword attended a banquet and adjusted in information seeking processes, a layer can be formed The limited reticular structure of number.Therefore during information service, as soon as when attend a banquet clearly input keyword when, can judge from the background The enterprise once provided by selection.Therefore, can every time attend a banquet scan for when, those are triggered by this keyword, The company information for once providing satisfactory service for user is arranged in front, has saved the information that user searches related service enterprise.

Fig. 7 shows the flow chart of the searching method of another embodiment of the invention.As shown in fig. 7, this method specifically includes that

Step S701, user inputs keyword or problem.

Step S702, match calculating for inputting with semantic net node.

Matching primitives can be carried out according to the expression for the semantic net and user's input that above-mentioned search engine implementation is formed, Matching primitives can also be carried out according to the semantic net that artificial statistics is formed.In one embodiment, step S711, search semanteme are executed Gateway keyword, information representation library, according in the semantic net keyword carry out matching primitives, judge user input expression whether be One normal expression.

In one embodiment, if user's input is an expression,

Step S703, judge whether to reach evaluation threshold value.

If not up to threshold value, step S712, into search routine.

Step S705, semantic net start node is selected.

Step S706, the weight distance that start node broadcasts information to target is calculated.

Step S707, semantic net keyword, information representation library.Inquire the semantic net keyword, whether information representation library deposits In recommended keywords and information on services, enter search routine if not finding and thening follow the steps S712, if finding, executes step S708。

Step S712, into search routine.

When user only selects the keyword recommended, there is no after the objective result of selection recommendation, search system needs root According to the search click commands of user, search in the database again.Search result at this time is arranged according to system relatedness computation Algorithm, the correlation that semantic net can be accumulated as general search system, commonly search by an input parameter of relevancy algorithm Search system of the cable system for the search system based on semantic net, can refer to it is existing be not based on semantic net be System.

Step S708, recommend the semantic net keyword and corresponding objective result on threshold value.

Step S709, user judges recommendation results and selects.

It can be matched using the method for similarity with keyword in semantic net.If the expression of user's input is one Word then directly carries out matching primitives with the keyword in semantic net, according to the keyword of user's input and semantic net keyword Similarity determines the start node of search.If user's input is a sentence, segmented, is determined according to certain segmenting method User inputs multiple keywords in expression, multiple start nodes can be selected in semantic net, according to node each in semantic net The weight distance between objective result, for user recommend objective result.

Step S713, record user selection using negative-feedback adjusting and optimizing semantic net, and enters search routine.

After user inputs expression, according to the objective result of semantic net recommendation or user oneself selection with negative feedback mode Adjust above-mentioned semantic net.

Step S710, record user selects and optimizes semantic net.

Searching method provided in an embodiment of the present invention be the searching request that a kind of Kernel-based methods knowledge concepts extract understand and As a result the method recommended, while mark prompt, search system can be understood based on the searching request of domain knowledge and service-aware It can prompt the user on how to correctly enter letter according to the information content feature that backstage can service when user scans for input Breath, or feedback arrives user interface at the first time backstage important service information.

The variation of the Information Requirement Characteristic and customer information requirement environment of the searching method perception user of the embodiment of the present invention, The sightless integrity mark of user is carried out to the searching request of user, the search result of search engine is made to be more in line with user's Demand, and adapt to the requirement of demand environment.

Fig. 8 shows the schematic diagram of the semantic net of a sphere of learning one professor group of one embodiment of the invention, is based on the language Adopted net, if it includes multiple keywords such as " Di Mubainasi-Lee ", " Di Muli ", " Bai Na that 1 node is taught in semantic net Si-Lee ", " Tim Lee ", " Berners-Lee " " Tim Berners-Lee ", if user inputs the name such as the " base of a fruit of professor 1 Mu Baina-Lee ", according to the semantic net, the keyword that search system inputs user and the keyword phase in semantic net dictionary The word of matching, discovery user's input is alike compared with " Di Mubainasi-Lee ", therefore can choose the professor 1 in semantic net The node at place is start node, provides information on services according to the objective result being connected with the start node for user.

Professor 1 is in the endpoint node of semantic net, is connected directly with objective result, has connection relationship with the start node Objective result includes " university " " professor " " author " " semantic net " " computer network system ", if professor 1 and above-mentioned objective result Weight distance be " 100 " " 200 " " 300 " " 400 " " 1000 ", then according to weight distance be user recommendation objective result row Sequence is " computer network system " " semantic net " " author " " professor " " university ".Prompt keyword can also be provided for user such as " professor 2 " " field 1 " " works 1 " " works 3 " " works 2 " " student 1 " " laboratory 1 ".Comprehensively consider the relationship of path and weight, Each multi-stress for calculating each keyword and associated objective result in the semantic net, the multi-stress according to weight with It is obtained apart from aggregative weighted.If after judging start node, according to the multi-stress size of the start node and each objective result Directly recommend objective result or prompt keyword for user.The specific method for recommending objective result or the method for prompting keyword Identical as the recommended method in above-mentioned searching method, details are not described herein again.

The searching method of the embodiment of the present invention is related to being directed in the information searches inquiry system service processes such as search engine, Evaluation that is personal for specific user, answering using the problems in groups of users, specific vertical industry service process, can accumulate Extraction and knowledge accumulation of application message etc..It can be according in service process, for the record that particular problem is answered, by similar Answer is answered, obtains the initiation problem for triggering this answer in turn.Further, similar initiation problem is inputted in user Afterwards, it is not required to the adjustment by user's keyword, directly provides possible search answer for user.

Fig. 9 shows the structural schematic diagram of the search engine system of one embodiment of the invention.As shown in figure 9, search engine It realizes and the module of optimization side includes:

User's usage data record module 905 records keyword and selection target result information, for example, record one fixed During the information service of justice, keyword, the keyword of adjustment and the search result of final choice of user's input.

Serve log library 906, the information that storage user's usage data record module 905 is recorded.

User's use information analysis evaluation module 907 inputs keyword and keyword criteria library Zhong Guan for evaluating user The similarity of keyword and semantic net node keyword.

Candidate keywords, problem add up library 908, the keyword of user's storage standards, and the keyword in this dictionary is available α is calculated to evaluate₁。

Keyword, information representation library 909, keyword and objective result based on user count keyword, information representation Library.

Semantic net forms and optimizes computing module 910 and forms a reasoning semantic net, the language according to user's search behavior Adopted net is constantly optimised during Reusability.

Recommending module 911 based on semantic net, the objective result for selecting the keyword recommended or selection to recommend, gives user Certain search is prompted, and after a user input, is searched by understanding that the information Perception user for analyzing user's input is possible Hitch fruit is prompted to user.

User's application side includes following module:

User inputs keyword module 901, for receiving the search key of user's input；

Information search recommending module 902 can be recommended based on semantic net information and be supplied to user.

Recommendation results display module 903, user, which applies, shows different contents according to different calculated result, such as recommends Keyword, recommended keywords are corresponding as a result, final search result etc..

Into search routine module 904, if the non-selected search engine implementation process of user recommend as a result, if enter it is common Search routine executes common search.

When user only select recommend keyword, there is no selection recommend search result after, search engine realize and Optimize side to need to be searched in the database again according to the search click commands of user, search result arrangement at this time is according to being The relatedness computation algorithm that system was set originally, the correlation of semantic net accumulation can be used as an input of system relevancy algorithm Parameter.

The search engine system of the embodiment of the present invention is not to substitute existing search system completely, and effect mainly has two A: one is the semantic net realized according to the input of user and search engine, recommends clothes relevant to input keyword for user It is engaged in information, services what successfully search objective result record COMPREHENSIVE CALCULATING went out by history when this relevant information on services.Separately One effect is to recommend currently to input relevant input keyword to user, correction user's mistake or not search result it is defeated Enter or provide the user with it is possible correctly enter, when user do not select these recommend keyword and search result when, be System can be inputted still according to the input of user.

For inputting the sequence of the search result carried out according to user, the accumulation adjustment sequence for considering semantic net can choose As a result, can not also consider the accumulation of semantic net, it is ranked up according to the relatedness computation method of system itself.

The search engine system of the embodiment of the present invention, according to for a certain individual, groups of users, field or a period of time Interior information on services searches for record, and study user searches for use habit, refines a kind of keyword, information representation mode and final mesh Mark the semanteme and conceptual relation between result；Conceptual understanding is carried out to the information that backstage can service simultaneously, is extracted, it is defeated to form user Enter the corresponding concept characteristic network between final service result.When user inputs a searching request, drawn according to this search The method or apparatus of system offer is held up, after recommending information on services expression relevant to this searching request for user and can service Station information.It is possible to further realize the concept extraction and the application that are directed to search engine history service data, one is provided for user Kind input method, information cuing method or information recommendation system.

Figure 10 shows the structural block diagram of the search engine realization device of one embodiment of the invention, and as shown in Figure 10, this is searched Index holds up realization device 1000 and includes:

Module 1001 is obtained, for obtaining the search expression history and selection that input in user one continuous search process Objective result；

Determining module 1002, for determining expression keyword sequence according to search expression history；

Generation module 1003, for including in semantic net according to keyword sequence and objective result generative semantics net is expressed Express the connection relationship and weight between keyword and between expression keyword and objective result.

The search engine realization device of the embodiment of the present invention, the target knot of expression keyword and selection based on user's input Fruit carries out statistics and establishes semantic net, obtains between keyword and keyword and the distance between keyword and objective result and weighs Series of fortified passes system provides service based on the semantic net in user's search process for user, can save user and search for information time, More acurrate efficient information service is provided for user.

In one embodiment, obtain module be used for obtain user input expression keyword, adjustment expression keyword with And the objective result selected according to search result.

Figure 11 shows the structural block diagram of the search engine realization device of another embodiment of the present invention, as shown in figure 11, should Search engine realization device 1100 includes: to obtain module 1001, determining module 1002, generation module 1003；And

Service field determining module 1104, for determining service field, service provider includes sole user, group；

Information service process judgment module 1105, for judging under service field in the expression keyword sequence of user's input Express keyword and according to expression Keyword Selection objective result process whether be complete information service process.

In one embodiment, the search engine realization device 1100 further include: evaluation module 1106, if being inputted for user Expression keyword be unsatisfactory for semantic web updates rule, then cast out the expression keyword；

If the expression keyword that evaluation module 1106 is also used to user's input meets semantic web updates rule, user's input The similarity for expressing keyword in keyword and semantic net is less than similarity threshold, closes in the expression of semantic net addition user's input Keyword as the new keywords in semantic net, and record user selection objective result, generate or update the new keywords with The multi-stress of connection relationship weight and the new keywords and objective result in semantic net between other keywords；

If the expression keyword that evaluation module 1106 is also used to user's input meets semantic web updates rule, user's input The similarity for expressing keyword in keyword and semantic net is more than or equal to similarity threshold, keyword in generation or update semantics net With the multi-stress of objective result, multi-stress be according to the shortest path factor of the keyword in semantic net to objective result with And weight factor determination, weight factor is the number normalizing that objective result is reached according to keyword in the semantic net of historical statistics Change determination.

In one embodiment, information service process judgment module 1105 is used for:

Determine that user inputs number of clicks of the process to each objective result of expression key word search targets result；If target As a result number of clicks is greater than information service threshold value, then service process is determined as complete service process, or user is inputted The process of the objective result selected after expression keyword is determined as complete service process.

Figure 12 shows the structural block diagram of the searcher of one embodiment of the invention, as shown in figure 12, the searcher 1200 include:

Key word analysis module 1201 is expressed, the expression keyword inputted in expression formula for determining user；

Start node analysis module 1202, for determining expression keyword matched start node in semantic net, wherein Semantic net based in user's search history expression keyword and objective result generate；

Determining module 1203, for determining objective result according to start node in semantic net, objective result is according to initial The distance and weight of node to objective result determine；

Cue module 1204, for prompting objective result.

The searcher of the embodiment of the present invention, the expression query semantics net that can be inputted according to user, according to the defeated of user Enter to be expressed as user and prompt keyword and information service, user can be saved and search for information time, provide more acurrate height for user The information service of effect.

In one embodiment, determining module is also used in semantic net determine prompt keyword according to start node；

Cue module is also used to prompt prompt keyword.

In one embodiment, cue module is used for:

If start node is not the endpoint node of semantic net, selects start node to develop to the path of objective result, determine Node on path is as prompt keyword.

In one embodiment, cue module, it is if being the endpoint node of semantic net for start node, start node is corresponding Objective result be prompted to user；

And/or

If cue module is used for the endpoint node that start node is not semantic net, the weight factor of selection target result is greater than The objective result of weight threshold；It selects start node to develop to the path factor of objective result and is greater than the path of path threshold；Really Other nodes on path are determined as prompt keyword；Prompt prompt keyword and/or multiple objective results, wherein semantic net In node be the same or similar crucial phrase at set, weight factor be according to keyword in the semantic net of historical statistics Reach the number normalization determination of objective result.

Figure 13 shows the structural block diagram of the search engine realization device of another embodiment of the invention.Search engine is real Existing device 1300 can be the host server for having computing capability, personal computer PC or portable portable computing Machine or terminal etc..The specific embodiment of the invention does not limit the specific implementation of calculate node.

Search engine realization device 1300 includes processor (processor) 1310, communication interface (Communications Interface) 1320, memory (memory) 1330 and bus 1340.Wherein, processor 1310, communication interface 1320 and Memory 1330 completes mutual communication by bus 1340.

Communication interface 1320 is used for and network device communications, and wherein the network equipment includes such as Virtual Machine Manager center, is total to Enjoy storage etc..

Processor 1310 is for executing program.Processor 1310 may be a central processor CPU or dedicated collection At circuit ASIC (Application Specific Integrated Circuit), or it is arranged to implement the present invention One or more integrated circuits of embodiment.

Memory 1330 is for storing file.Memory 1330 may include high speed RAM memory, it is also possible to further include non- Volatile memory (non-volatile memory), for example, at least a magnetic disk storage.Memory 1330 is also possible to deposit Memory array.Memory 1330 is also possible to by piecemeal, and block can be combined into virtual volume by certain rule.

In a kind of possible embodiment, above procedure can be the program code for including computer operation instruction.The journey Sequence is particularly used in: obtaining the objective result of the search expression history and selection that input in user one continuous search process； Expression keyword sequence is determined according to search expression history；According to expression keyword sequence and objective result generative semantics net, It include the connection relationship and weight expressed between keyword and between expression keyword and objective result in semantic net.

In one embodiment, the search expression history and selection inputted in user one continuous search process is obtained Objective result includes:

Obtain the expression keyword of user's input, the expression keyword of adjustment and the target knot selected according to search result Fruit.

In one embodiment, before according to expression keyword sequence and objective result generative semantics net further include:

Determine service field, service provider includes sole user, group；

Judge to express keyword under service field in the expression keyword sequence of user's input and according to the crucial selected ci poem of expression Whether the process for selecting objective result is complete information service process.

It in one embodiment, include between expression keyword and between expression keyword and objective result in semantic net Connection relationship and weight, comprising:

If the expression keyword of user's input is unsatisfactory for semantic web updates rule, cast out the expression keyword；

If the expression keyword of user's input meets semantic web updates rule, the expression keyword and semantic net of user's input The similarity of middle keyword is less than similarity threshold, in the expression keyword that semantic net addition user inputs as in semantic net New keywords, and the objective result of user's selection is recorded, generate or update other keywords in the new keywords and semantic net Between connection relationship weight and the new keywords and objective result multi-stress；

If the expression keyword of user's input meets semantic web updates rule, the expression keyword and semantic net of user's input The similarity of middle keyword be more than or equal to similarity threshold, generate or update semantics net in the synthesis of keyword and objective result because Son, multi-stress are determined according to the shortest path factor and weight factor of the keyword in semantic net to objective result, Weight factor is that the number normalization determination of objective result is reached according to keyword in the semantic net of historical statistics.

In one embodiment, judge to express keyword and root in the expression keyword sequence of user's input under service field Whether the process according to expression Keyword Selection objective result is complete information service process, comprising: determines that user inputs expression Number of clicks of the process of key word search targets result to each objective result；If the number of clicks of objective result takes greater than information Service process is then determined as complete service process, or user is inputted to the target knot selected after expression keyword by business threshold value The process of fruit is determined as complete service process.

The search engine realization device of the embodiment of the present invention is carried out according to the expression formula of user's input and selection target result Statistics establishes semantic net, provides service based on the semantic net for user, can save user and search for information time, provide for user More acurrate efficient information service.

Figure 14 shows the structural block diagram of the searcher of another embodiment of the invention.Searcher 1400 can be Host server, personal computer PC or the portable portable computer or terminal etc. for having computing capability.The present invention Specific embodiment does not limit the specific implementation of calculate node.

Searcher 1400 includes processor (processor) 1410, communication interface (Communications Interface) 1420, memory (memory) 1430 and bus 1440.Wherein, processor 1410, communication interface 1420 and Memory 1430 completes mutual communication by bus 1440.

Communication interface 1420 is used for and network device communications, and wherein the network equipment includes such as Virtual Machine Manager center, is total to Enjoy storage etc..

Processor 1410 is for executing program.Processor 1410 may be a central processor CPU or dedicated collection At circuit ASIC (Application Specific Integrated Circuit), or it is arranged to implement the present invention One or more integrated circuits of embodiment.

Memory 1430 is for storing file.Memory 1430 may include high speed RAM memory, it is also possible to further include non- Volatile memory (non-volatile memory), for example, at least a magnetic disk storage.Memory 1430 is also possible to deposit Memory array.Memory 1430 is also possible to by piecemeal, and block can be combined into virtual volume by certain rule.

In a kind of possible embodiment, above procedure can be the program code for including computer operation instruction.The journey Sequence is particularly used in: determining the expression keyword that user inputs in expression formula；Determine that expression keyword is matched in semantic net Start node, wherein semantic net based in user's search history expression keyword and objective result generate；The root in semantic net Determine that objective result, objective result are determined according to the distance and weight of start node to objective result according to start node；Prompt mesh Mark result.

In one embodiment, further includes: prompt keyword is determined according to start node in semantic net；Prompt prompt is closed Keyword.

In one embodiment, if in semantic net according to start node determine prompt keyword include: start node not For the endpoint node of semantic net, start node is selected to develop to the path of objective result, determines the node on path as prompt Keyword.

In one embodiment, if start node is the endpoint node of semantic net, by the corresponding objective result of start node It is prompted to user；If start node is not the endpoint node of semantic net, the weight factor of selection target result is greater than weight threshold Objective result；It selects start node to develop to the path factor of objective result and is greater than the path of path threshold；It determines on path Other nodes as prompt keyword；Prompt prompt keyword and/or multiple objective results；Wherein, the node in semantic net For the same or similar crucial phrase at set, weight factor be according in the semantic net of historical statistics keyword reach target As a result number normalization determination.

The searcher of the embodiment of the present invention, the expression query semantics net that can be inputted according to user, according to the defeated of user Enter to be expressed as user and prompt keyword and information service, user can be saved and search for information time, provide more acurrate height for user The information service of effect.

Description of the invention is given for the purpose of illustration and description, and is not exhaustively or will be of the invention It is limited to disclosed form.Many modifications and variations are obvious for the ordinary skill in the art.It selects and retouches It states embodiment and is to more preferably illustrate the principle of the present invention and practical application, and those skilled in the art is enable to manage The solution present invention is to design various embodiments suitable for specific applications with various modifications.

Claims (10)

1. a kind of search engine implementation method characterized by comprising

Obtain the objective result of the search expression history and selection that input in user one continuous search process；

Expression keyword sequence is determined according to described search expression formula history；

According to the expression keyword sequence and the objective result generative semantics net to be used for search engine；

Wherein, if the expression keyword of user's input is unsatisfactory for semantic web updates rule, cast out the expression keyword；

If the expression keyword of user's input meets semantic web updates rule, closed in the expression keyword and semantic net of user's input The similarity of keyword is less than similarity threshold, adds the expression keyword of user's input as semantic net in the semantic net In new keywords, and record the objective result of user selection, generate or update its in the new keywords and semantic net The multi-stress of connection relationship weight and the new keywords and objective result between his keyword；

If the expression keyword of user's input meets semantic web updates rule, closed in the expression keyword and semantic net of user's input The similarity of keyword is more than or equal to similarity threshold, generate or update in the semantic net synthesis of keyword and objective result because Son, the multi-stress are the shortest path factor and weight factor according to the keyword in semantic net to objective result Determining, weight factor is that the number normalization determination of objective result is reached according to keyword in the semantic net of historical statistics.

2. the method according to claim 1, wherein obtaining the search inputted in user one continuous search process The objective result of expression formula history and selection includes:

Obtain the initial expression keyword of user's input, the expression keyword of adjustment and the target knot selected according to search result Fruit.

3. the method according to claim 1, wherein according to the expression keyword sequence and the objective result Before generative semantics net further include:

Determine that service field, the service field include sole user, group, vertical field or integrated information service field；

Judge to express keyword under the service field in the expression keyword sequence of user's input and according to the expression key Whether the process that selected ci poem selects objective result is complete information service process.

4. according to the method described in claim 3, it is characterized in that, judging that the expression that user inputs under the service field is crucial In word sequence express keyword and according to it is described expression Keyword Selection objective result process whether be complete information service Process, comprising:

Determine that user inputs number of clicks of the service process to each objective result of expression key word search targets result；If target As a result number of clicks is greater than information service threshold value, then the service process is determined as complete service process；

And/or

The process that user inputs the objective result selected after expression keyword is determined as complete service process.

5. a kind of searching method characterized by comprising

Determine the expression keyword that user inputs in expression formula；

Determine expression keyword matched start node in semantic net, wherein the semantic net is based on expression keyword Sequence and the objective result of selection generate, and the expression keyword sequence is inputted according in user one continuous search process What search expression history determined；

Determine that objective result, the objective result are arrived according to the start node according to the start node in the semantic net The distance and weight of the objective result determine；

Prompt the objective result；

Wherein, if the start node is the endpoint node of the semantic net, the corresponding objective result of the start node is mentioned Show to user；

If the start node is not the endpoint node of the semantic net, the weight factor of selection target result is greater than weight threshold Objective result；It selects the start node to develop to the path factor of the objective result and is greater than the path of path threshold；Really Other nodes on the fixed path are as prompt keyword；Prompt the prompt keyword and/or the multiple objective result；

Wherein, the node in the semantic net be the same or similar crucial phrase at set, weight factor be according to history Keyword reaches the number normalization determination of objective result in the semantic net of statistics.

6. a kind of search engine realization device characterized by comprising

Module is obtained, for obtaining the target knot of the search expression history and selection that input in user one continuous search process Fruit；

Determining module, for determining expression keyword sequence according to described search expression formula history；

Generation module, for according to the expression keyword sequence and the objective result generative semantics net, in the semantic net Including the connection relationship and weight between the expression keyword and between the expression keyword and the objective result；

Evaluation module is cast out the expression and is closed if the expression keyword for user's input is unsatisfactory for semantic web updates rule Keyword；

The expression keyword that the evaluation module is also used to user's input meets semantic web updates rule, and the expression of user's input is closed The similarity of keyword is less than similarity threshold in keyword and semantic net, adds the expression of user's input in the semantic net Keyword is as the new keywords in semantic net, and the objective result for recording user's selection generates or update the new key The multi-stress of connection relationship weight and the new keywords and objective result in word and semantic net between other keywords；

If the expression keyword that the evaluation module is also used to user's input meets semantic web updates rule, the expression of user's input The similarity of keyword is more than or equal to similarity threshold in keyword and semantic net, generates or update keyword in the semantic net With the multi-stress of objective result, the multi-stress is the shortest path according to the keyword in semantic net to objective result What the diameter factor and weight factor determined, weight factor is to reach objective result according to keyword in the semantic net of historical statistics Number normalization determination.

7. device according to claim 6, which is characterized in that

The module that obtains is used to obtain the expression keyword of user's input, the expression keyword of adjustment and according to search result The objective result of selection.

8. device according to claim 6, which is characterized in that further include:

Service field determining module, for determining service field, the service field include sole user, group, vertical field, Or integrated information service field；

Information service process judgment module is expressed in the expression keyword sequence of user's input under the service field for judging Keyword and according to it is described expression Keyword Selection objective result process whether be complete information service process.

9. device according to claim 8, which is characterized in that the information service process judgment module is used for:

Determine that user inputs number of clicks of the process to each objective result of expression key word search targets result；If objective result Number of clicks be greater than information service threshold value, then the service process is determined as complete service process；

And/or

The process that user inputs the objective result selected after expression keyword is determined as complete service process.

10. a kind of search engine device characterized by comprising

Key word analysis module is expressed, the expression keyword inputted in expression formula for determining user；

Start node analysis module, for determining expression keyword matched start node in semantic net, wherein described Semantic net is generated based on the objective result of expression keyword sequence and selection, and the expression keyword sequence is according to user one What the search expression history inputted in continuous search process determined；

Determining module, for determining objective result according to the start node in the semantic net, the objective result according to The distance and weight of the start node to the objective result determine；

Cue module, for prompting the objective result；

Wherein, the cue module, if being the endpoint node of the semantic net for the start node, by the start node Corresponding objective result is prompted to user；

And/or

If the cue module is used for the endpoint node that the start node is not the semantic net, the weight of selection target result The factor is greater than the objective result of weight threshold；It selects the start node to develop and is greater than road to the path factor of the objective result The path of diameter threshold value；Determine other nodes on the path as prompt keyword；Prompt the prompt keyword and/or institute State multiple objective results；

Wherein, the node in the semantic net be the same or similar crucial phrase at set, weight factor is according to history Keyword reaches the number normalization determination of objective result in the semantic net of statistics.