Connect public, paid and private patent data with Google Patents Public Datasets

Search method and search engine

Info

Publication number
CN102096717A
CN102096717A CN 201110038433 CN201110038433A CN102096717A CN 102096717 A CN102096717 A CN 102096717A CN 201110038433 CN201110038433 CN 201110038433 CN 201110038433 A CN201110038433 A CN 201110038433A CN 102096717 A CN102096717 A CN 102096717A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
search
query
command
method
intention
Prior art date
Application number
CN 201110038433
Other languages
Chinese (zh)
Other versions
CN102096717B (en )
Inventor
刘建柱
Original Assignee
百度在线网络技术(北京)有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Abstract

The invention provides a search method. The search method comprises the following steps of: S1, receiving a query command; S2, performing demand intention analysis on the query command based on a knowledge base, and defining the demand intention of the query command; S3, searching the query command carrying the demand intention in a database to obtain a search result; and S4, outputting the search result. Compared with the prior art, the search method has the advantages that: on the basis of the knowledge base, the query command input by a user is understood better; the intention of the query command is analyzed; the structure of the query command is analyzed to perform semantic content expansion on the query command so as to better guide a search engine to select quality resources to meet the search requirements on the user; therefore, the search efficiency of the user is improved and the network traffic is saved.

Description

搜索方法及搜索引擎 Search method and search engine

技术领域 FIELD

[0001] 本发明涉及搜索引擎技术,尤其涉及一种基于知识库对查询指令进行需求分析与解析的搜索方法及搜索引擎。 [0001] The present invention relates to a search engine technology, particularly to a database based on the query instructions needs analysis and parsing search method and a search engine.

背景技术 Background technique

[0002] 随着互联网上信息的飞速增长,网络上充斥了越来越多的冗余信息,而对于在网络上搜寻自己所需要信息的互联网用户而言,面对这些漫无边际的信息无疑像大海捞针。 [0002] With the rapid growth of information on the Internet, more and more full of redundant information on the network, and search for the information they need in terms of Internet users on the network, no doubt to face these endless information like needle in a haystack . 搜索引擎的出现无疑在一定程度上为用户的搜索需求带来了很大便利。 Occurrences of the search engine is undoubtedly a certain extent, the search needs of users has brought great convenience. 搜索引擎是一种在网络上应用的软件系统,其以一定的策略在网络上搜集和发现信息,并在对信息进行处理和组织后,为用户提供互联网上的信息搜索服务。 Search engines are software system in use on the network, which to a certain strategy to gather and find information on the Web, and after processing the information and organization, to provide users with information search services on the Internet. 通常,这种软件系统提供一个网页界面, 让用户在客户端通过浏览器软件提交搜索词,然后很快返回一个可能和用户输入的搜索内容相关的信息列表。 Often, this software provides a web interface, allowing users to submit search terms by the client browser software, then quickly returns a list of information a user may enter and search for content related. 这个列表通常会包括上万个条目,每个条目代表一篇搜索到的相关网页。 This list usually includes tens of thousands of entries, each entry represents a search for the relevant pages.

[0003] 过去十几年以来,相应地,众多的互联网搜索引擎及对应的网站应运而生,这中间的佼佼者包括百度公司的百度搜索(WWW. baidu. com)和谷歌公司的谷歌搜索(www. google, cn)。 [0003] Over the past ten years, accordingly, a large number of Internet search engines and the corresponding website came into being, which is the middle of the crowd, including Baidu search company Baidu (WWW. Baidu. Com) and Google's Google search ( www. google, cn).

[0004] 现有的搜索引擎对用户输入的查询指令大多是基于查询指令字符理解的,例如,用户输入查询指令为“Nokia手机”,基于现有的搜索引擎只能将该查询指令分词为“Nokia”和“手机”,且通过该分词结果在网页数据库索引中进行检索,将文本包括“Nokia” 和“手机”的网页Url输入,形成搜索结果,然而这种搜索引擎并不能对用户的查询指令进行内容与语义层次上的理解,例如,用户输入查询指令为“Nokia手机”,其并不能将这个查询指令理解为“Nokia”为“手机”中的一种品牌;当然,更不能理解查询指令的需求意图,以及查询指令的结构,不能对查询指令进行语义内容扩充等。 [0004] instruction existing search engine query entered by the user are mostly based on the query command character to understand, for example, a user enters a query command is "Nokia mobile phone", only the query instruction word based on existing search engines " Nokia "and" mobile phone ", and the results through the word on the page database index search, the text includes" Nokia "and" mobile phone "page Url input, the formation of the search results, however, this search engine does not query the user instructions on the content and understand the semantic level, for example, a user enters a query command is "Nokia mobile phone", it does not bring this query command understood as "Nokia" kind of brand "mobile phone" in the range; of course, can not understand the query demand intent instruction, as well as query command structure, you can not query instructions semantic content expansion and so on. 对于用户输入的表达形式多样化、需求意图多样化的查询指令,现有的基于字符的搜索引擎已经不能更好的满足用户的需求,造成用于查找不全,需要多次输入不同的查询指令才可能找到需要的搜索结果,搜索效率较低,浪费网络资源的问题。 For the expression of diverse forms of user input, diverse needs of intent query instruction, based on the existing character of the search engines can not better meet the needs of users, find the cause for failure, you need to repeatedly enter different query commands only You may need to find the search results, search for low efficiency, waste of network resources.

发明内容 SUMMARY

[0005] 本发明的目的在于提供一种改进的搜索方法,其可在知识库的基础上,更好的理解用户输入的查询指令,分析查询指令的以图,解析查询指令的结构,对查询指令进行语义内容扩充。 [0005] The object of the present invention to provide an improved search method, which may be based on the knowledge base, a better understanding of query instruction input by a user, an instruction to analyze the query FIG resolution structure of the query instruction, the query instructions semantic content expansion.

[0006] 本发明的目的还在于提供一种实现上述搜索方法的改进的搜索引擎。 [0006] The object of the present invention is to provide an improved above-described search method implemented in a search engine.

[0007] 为实现上述发明目的之一,本发明第一实施方式提供一种搜索方法,包括以下步骤: [0007] In order to achieve the above object, one embodiment of a first embodiment of the present invention provides a searching method, comprising the steps of:

[0008] Si、接收查询指令; [0008] Si, receiving a query command;

[0009] S2、基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图; [0009] S2, knowledge of the inquiry instruction needs analysis based on the intent, clearly the intent of the query needs instruction;

[0010] S3、将带有需求意图的所述查询指令在数据库中搜索,得到搜索结果; [0010] S3, the intent with the needs of the database search query instruction, obtain search results;

[0011] S4、输出所述搜索结果。 [0011] S4, outputs the search result.

[0012] 作为本发明的进一步改进,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 [0012] As a further improvement of the present invention, the database is a web page or store the corresponding demand intended vertical search database.

[0013] 作为本发明的进一步改进,在所述S2步骤和S3步骤间,还包括语义扩充步骤: [0013] As a further improvement of the present invention, between step S2 and the step S3, further comprising the step of semantic extension:

[0014] 基于所述知识库对所述查询指令进行语义扩充。 [0014] semantic query expansion of the instruction based on the knowledge.

[0015] 作为本发明的进一步改进,所述“基于知识库对所述查询指令进行需求意图分析, 明确所述查询指令的需求意图”具体包括以下流程: [0015] As a further improvement of the present invention, the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes:

[0016] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分; [0016] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring;

[0017] S201、将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段; [0017] S201, the query command and the matching pieces of knowledge, to give at least a fragment of the query instruction information matches;

[0018] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; [0018] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction;

[0019] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分; [0019] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score knowledge base;

[0020] S204、判断所述知识库整体需求得分是否大于一设定阈值; [0020] S204, determines whether or not the overall demand knowledge score greater than a predetermined threshold value;

[0021] S205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图; [0021] S205, if greater than the set threshold value, places the highest scoring overall demand knowledge of the demand type as a query instruction needs intention;

[0022] S206、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 [0022] S206, if less than the set threshold value, it is determined that the query instruction needs no intent.

[0023] 作为本发明的进一步改进,所述“基于知识库对所述查询指令进行需求意图分析, 明确所述查询指令的需求意图”具体包括以下流程: [0023] As a further improvement of the present invention, the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes:

[0024] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分; [0024] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring;

[0025] S201、将所述查询指令与知识片段和表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一表达模板; [0025] S201, the query instructions and pieces of knowledge and expression template matching, and obtain at least one fragment of a knowledge of the expression template match query instruction;

[0026] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; [0026] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction;

[0027] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分; [0027] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score knowledge base;

[0028] S204、对所述查询指令在表达模板层面上进行打分,得到表达模板得分; [0028] S204, the query instruction template scoring in expression levels, resulting in the expression template score;

[0029] S205、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分; [0029] S205, the overall demand knowledge score and the score of the template expression intensity score as a weighted sum query instruction needs;

[0030] S206、判断所述查询指令需求强度得分是否大于一设定阈值; [0030] S206, determines whether the query instruction needs intensity score greater than a predetermined threshold value;

[0031] S207、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图; [0031] S207, if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intention;

[0032] S208、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 [0032] S208, if less than the set threshold value, it is determined that the query instruction needs no intent.

[0033] 为实现上述发明目的之一,本发明第二实施方式提供一种搜索方法,包括以下步骤: [0033] In order to achieve the above object, one embodiment of a second embodiment of the present invention provides a searching method, comprising the steps of:

[0034] Si、接收查询指令; [0034] Si, receiving a query command;

[0035] S2、基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图,同时,基于所述知识库对所述查询指令进行语义扩充; [0035] S2, based on knowledge of the intention of the query instruction needs analysis, a clear intention of the inquiry demand command, while based on the knowledge of the semantic query expansion instruction;

[0036] S3、将带有需求意图并扩充语义的查询指令在数据库中搜索,得到搜索结果; [0036] S3, with the intention to demand and expand the semantic query command search the database to obtain search results;

[0037] S4、输出所述搜索结果。 [0037] S4, outputs the search result.

[0038] 作为本发明的进一步改进,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 [0038] As a further improvement of the present invention, the database is a web page or store the corresponding demand intended vertical search database.

[0039] 作为本发明的进一步改进,所述“基于知识库对所述查询指令进行需求意图分析, 明确所述查询指令的需求意图”具体包括以下流程: [0039] As a further improvement of the present invention, the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes:

[0040] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分; [0040] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring;

[0041] S201、将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段; [0041] S201, the query command and the matching pieces of knowledge, to give at least a fragment of the query instruction information matches;

[0042] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; [0042] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction;

[0043] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分; [0043] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score knowledge base;

[0044] S204、判断所述知识库整体需求得分是否大于一设定阈值; [0044] S204, determines whether or not the overall demand knowledge score greater than a predetermined threshold value;

[0045] S205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图; [0045] S205, if greater than the set threshold value, places the highest scoring overall demand knowledge of the demand type as a query instruction needs intention;

[0046] S206、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 [0046] S206, if less than the set threshold value, it is determined that the query instruction needs no intent.

[0047] 作为本发明的进一步改进,所述“基于知识库对所述查询指令进行需求意图分析, 明确所述查询指令的需求意图”具体包括以下流程: [0047] As a further improvement of the present invention, the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes:

[0048] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分; [0048] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring;

[0049] S201、将所述查询指令与知识片段和表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一表达模板; [0049] S201, the query instructions and pieces of knowledge and expression template matching, and obtain at least one fragment of a knowledge of the expression template match query instruction;

[0050] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; [0050] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction;

[0051] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分; [0051] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score knowledge base;

[0052] S204、对所述查询指令在表达模板层面上进行打分,得到表达模板得分; [0052] S204, the query instruction template scoring in expression levels, resulting in the expression template score;

[0053] S205、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分; [0053] S205, the overall demand knowledge score and the score of the template expression intensity score as a weighted sum query instruction needs;

[0054] S206、判断所述查询指令需求强度得分是否大于一设定阈值; [0054] S206, determines whether the query instruction needs intensity score greater than a predetermined threshold value;

[0055] S207、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图; [0055] S207, if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intention;

[0056] S208、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 [0056] S208, if less than the set threshold value, it is determined that the query instruction needs no intent. [0057] 为实现上述发明目的之一,本发明第三实施方式提供一种搜索方法,包括以下步骤: [0057] In order to achieve the above object, one embodiment of a third embodiment of the present invention provides a searching method, comprising the steps of:

[0058] Si、接收查询指令; [0058] Si, receiving a query command;

[0059] S2、基于知识库和表达模板库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图; [0059] S2, knowledge and expression to the query template library instruction needs analysis based on the intent, clearly the intent of the query needs instruction;

[0060] S3、将带有需求意图的所述查询指令在数据库中搜索,得到搜索结果; [0060] S3, the intent with the needs of the database search query instruction, obtain search results;

[0061] S4、输出所述搜索结果。 [0061] S4, outputs the search result.

[0062] 作为本发明的进一步改进,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 [0062] As a further improvement of the present invention, the database is a web page or store the corresponding demand intended vertical search database.

[0063] 作为本发明的进一步改进,在所述S2步骤和S3步骤间,还包括语义扩充步骤: [0063] As a further improvement of the present invention, between step S2 and the step S3, further comprising the step of semantic extension:

[0064] 基于所述知识库对所述查询指令进行语义扩充。 [0064] semantic query expansion of the instruction based on the knowledge.

[0065] 作为本发明的进一步改进,所述“基于知识库对所述查询指令进行需求意图分析, 明确所述查询指令的需求意图”具体包括以下流程: [0065] As a further improvement of the present invention, the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes:

[0066] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分; [0066] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring;

[0067] S201、将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段; [0067] S201, the query command and the matching pieces of knowledge, to give at least a fragment of the query instruction information matches;

[0068] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; [0068] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction;

[0069] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分; [0069] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score knowledge base;

[0070] S204、判断所述知识库整体需求得分是否大于一设定阈值; [0070] S204, determines whether or not the overall demand knowledge score greater than a predetermined threshold value;

[0071] S205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图; [0071] S205, if greater than the set threshold value, places the highest scoring overall demand knowledge of the demand type as a query instruction needs intention;

[0072] S206、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 [0072] S206, if less than the set threshold value, it is determined that the query instruction needs no intent.

[0073] 作为本发明的进一步改进,所述“基于知识库对所述查询指令进行需求意图分析, 明确所述查询指令的需求意图”具体包括以下流程: [0073] As a further improvement of the present invention, the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes:

[0074] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分; [0074] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring;

[0075] S201、将所述查询指令与知识片段和表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一表达模板; [0075] S201, the query instructions and pieces of knowledge and expression template matching, and obtain at least one fragment of a knowledge of the expression template match query instruction;

[0076] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; [0076] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction;

[0077] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分; [0077] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score knowledge base;

[0078] S204、对所述查询指令在表达模板层面上进行打分,得到表达模板得分; [0078] S204, the query instruction template scoring in expression levels, resulting in the expression template score;

[0079] S205、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分; [0079] S205, the overall demand knowledge score and the score of the template expression intensity score as a weighted sum query instruction needs;

[0080] S206、判断所述查询指令需求强度得分是否大于一设定阈值; [0080] S206, determines whether the query instruction needs intensity score greater than a predetermined threshold value;

11[0081] S207、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图; 11 [0081] S207, if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intention;

[0082] S208、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 [0082] S208, if less than the set threshold value, it is determined that the query instruction needs no intent.

[0083] 作为本发明的进一步改进,所述表达模板库的构建方法,包括以下流程: [0083] As a further improvement of the present invention, the expression construct template library method, comprising the following processes:

[0084] S300、抽取在用户历史行为库中包含知识片段的查询指令; [0084] S300, the extraction query instruction contains pieces of knowledge in user behavior history database;

[0085] S301、将所述知识库片段替换成通用符号,生成候选表达模板; [0085] S301, the knowledge base into a common symbol replaced fragment, the expression of generating candidate template;

[0086] S302、统计生成的所述候选表达模板符合的知识库片段的数量; [0086] S302, the number of statistics generated knowledge base segment of the expression of the candidate template matching;

[0087] S303、判断所述数量是否大于设定阈值; [0087] S303, determines whether the number is greater than the set threshold;

[0088] S304、若大于设定阈值,则将所述候选表达模板作为表达模板,并存于数据库中, 生成表达模板库; [0088] S304, if greater than the set threshold, the template expression as the expression of the candidate template, exist in the database, generating an expression template library;

[0089] S305、若小于设定阈值,则舍弃所述候选表达模板。 [0089] S305, if less than a set threshold value, the expression of the candidate template is discarded.

[0090] 为实现上述发明目的之一,本发明第四实施方式提供一种搜索方法,包括以下步骤: [0090] In order to achieve the above object, one embodiment of a fourth embodiment of the present invention provides a searching method, comprising the steps of:

[0091] Si、接收查询指令; [0091] Si, receiving a query command;

[0092] S2、基于知识库和表达模板库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图,同时,基于所述知识库对接收到的查询指令进行语义扩充; [0092] S2, knowledge and expression to the query template library instruction needs analysis based on the intent, clearly the intent of the query needs instruction, while the knowledge-based query command received semantic expansion;

[0093] S3、将带有需求意图并扩充语义的查询指令在数据库中搜索,得到搜索结果; [0093] S3, with the intention to demand and expand the semantic query command search the database to obtain search results;

[0094] S4、输出所述搜索结果。 [0094] S4, outputs the search result.

[0095] 作为本发明的进一步改进,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 [0095] As a further improvement of the present invention, the database is a web page or store the corresponding demand intended vertical search database.

[0096] 作为本发明的进一步改进,所述“基于知识库对所述查询指令进行需求意图分析, 明确所述查询指令的需求意图”具体包括以下流程: [0096] As a further improvement of the present invention, the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes:

[0097] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分; [0097] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring;

[0098] S201、将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段; [0098] S201, the query command and the matching pieces of knowledge, to give at least a fragment of the query instruction information matches;

[0099] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; [0099] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction;

[0100] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分; [0100] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score knowledge base;

[0101] S204、判断所述知识库整体需求得分是否大于一设定阈值; [0101] S204, determines whether or not the overall demand knowledge score greater than a predetermined threshold value;

[0102] S205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图; [0102] S205, if greater than the set threshold value, places the highest scoring overall demand knowledge of the demand type as a query instruction needs intention;

[0103] S206、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 [0103] S206, if less than the set threshold value, it is determined that the query instruction needs no intent.

[0104] 作为本发明的进一步改进,所述“基于知识库对所述查询指令进行需求意图分析, 明确所述查询指令的需求意图”具体包括以下流程: [0104] As a further improvement of the present invention, the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes:

[0105] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分; [0105] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring;

[0106] S201、将所述查询指令与知识片段和表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一表达模板; [0106] S201, the query instructions and pieces of knowledge and expression template matching, and obtain at least one fragment of a knowledge of the expression template match query instruction;

[0107] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; [0107] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction;

[0108] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分; [0108] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score knowledge base;

[0109] S204、对所述查询指令在表达模板层面上进行打分,得到表达模板得分; [0109] S204, the query instruction template scoring in expression levels, resulting in the expression template score;

[0110] S205、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分; [0110] S205, the overall demand knowledge score and the score of the template expression intensity score as a weighted sum query instruction needs;

[0111] S206、判断所述查询指令需求强度得分是否大于一设定阈值; [0111] S206, determines whether the query instruction needs intensity score greater than a predetermined threshold value;

[0112] S207、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图; [0112] S207, if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intention;

[0113] S208、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 [0113] S208, if less than the set threshold value, it is determined that the query instruction needs no intent.

[0114] 作为本发明的进一步改进,所述表达模板库的构建方法,包括以下流程: [0114] As a further improvement of the present invention, the expression construct template library method, comprising the following processes:

[0115] S300、抽取在用户历史行为库中包含知识片段的查询指令; [0115] S300, the extraction query instruction contains pieces of knowledge in user behavior history database;

[0116] S301、将所述知识库片段替换成通用符号,生成候选表达模板; [0116] S301, the knowledge base into a common symbol replaced fragment, the expression of generating candidate template;

[0117] S302、统计生成的所述候选表达模板符合的知识库片段的数量; [0117] S302, the number of statistics generated knowledge base segment of the expression of the candidate template matching;

[0118] S303、判断所述数量是否大于设定阈值; [0118] S303, determines whether the number is greater than the set threshold;

[0119] S304、若大于设定阈值,则将所述候选表达模板作为表达模板,并存于数据库中, 生成表达模板库; [0119] S304, if greater than the set threshold, the template expression as the expression of the candidate template, exist in the database, generating an expression template library;

[0120] S305、若小于设定阈值,则舍弃所述候选表达模板。 [0120] S305, if less than a set threshold value, the expression of the candidate template is discarded.

[0121] 相应地,作为实现上述发明另一目的,本发明一实施方式提供一种搜索引擎,包括: [0121] Accordingly, as a further object to achieve the above-described invention, the present invention provides an embodiment of a search engine, comprising:

[0122] UI模块,用于接收查询指令,且所述UI模块还用于接收搜索模块返回的搜索结果,并将所述搜索结果拼装为结果页面后输出; [0122] UI module, configured to receive a query instruction, and the UI module is further configured to receive the search module returns the search results, and the search results page to the result output assembly;

[0123] 需求意图分析模块,用于基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图; [0123] Demand intent analysis module, based on knowledge of the intended query instruction needs analysis, clear the query requirement command is intended;

[0124] 搜索模块,用于将带有需求意图的所述查询指令在数据库中搜索,得到搜索结果; [0124] The search module, with the demand for the intended search the database query instruction, obtain search results;

[0125] 知识库,用于存储先验知识。 [0125] knowledge base for storing a priori knowledge.

[0126] 作为本发明的进一步改进,所述搜索引擎还包括: [0126] As a further improvement of the present invention, the search engine further comprises:

[0127] web服务模块,用于通过网络协议接收客户端发出的查询指令,并将所述查询指令转到所述UI模块,且所述web服务模块还用于接收所述UI模块返回的结果页面,并将所述结果页面返回至所述客户端。 [0127] web service module for receiving a query instruction sent by the client through the network protocol, and the query instruction to the UI module and the web service module is further configured to receive the results returned by the UI module page and the results page returned to the client.

[0128] 作为本发明的进一步改进,所述搜索引擎还包括: [0128] As a further improvement of the present invention, the search engine further comprises:

[0129] 用户历史行为库,用于存储用户历史搜索记录。 [0129] user behavior history database for storing user search history records.

[0130] 作为本发明的进一步改进,所述用户历史搜索记录包括:查询指令、查询次数,以及加权点击数。 [0130] As a further improvement of the present invention, the user history search record comprising: a query instructions, queries, and the weighting clicks.

[0131] 作为本发明的进一步改进,所述搜索引擎还包括: [0131] As a further improvement of the present invention, the search engine further comprises:

[0132] 表达模板挖掘模块,用于根据所述知识库中的知识片段和所述用户历史行为库中的用户历史查询指令,挖掘表达模板,并将所述表达模板存储于表达模板库; [0132] Expression template mining module configured to query instruction based on knowledge of the historical knowledge base of the segment and the user behavior history of the user database mining expression template, and the template stored in the expression of the expression template library;

[0133] 表达模板库,用于存储由所述表达模板挖掘模块挖掘出的表达模板。 [0133] The expression template database, the template expressed by the expression mined mining module for storing the template.

[0134] 作为本发明的进一步改进,所述搜索引擎还包括: [0134] As a further improvement of the present invention, the search engine further comprises:

[0135] 结构分类模块,用于基于所述知识库对所述查询指令进行语义扩充。 [0135] The classification module structure, based on the knowledge of the semantic query expansion instruction.

[0136] 作为本发明的进一步改进,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 [0136] As a further improvement of the present invention, the database is a web page or store the corresponding demand intended vertical search database.

[0137] 作为本发明的进一步改进,所述网页存储库用于存储网页数据和该网页数据的索引信息; [0137] As a further improvement of the present invention, the page index information repository for storing web page data and the page data;

[0138] 所述垂直搜索数据库用于存储特定类别数据和该特定类别数据的索引信息。 [0138] The vertical search database for storing a particular category of data and index data of the specific categories of information.

[0139] 与现有技术相比,本发明的有益效果是:在知识库的基础上,更好的理解用户输入的查询指令,分析查询指令的以图,解析查询指令的结构,对查询指令进行语义内容扩充, 从而更好的指导搜索引擎选择优质的资源满足用户的搜索需求,使得用户搜索效率提高, 节约网络流量。 [0139] Compared with the prior art, the beneficial effect of the invention is: on the basis of the knowledge base, a better understanding of query instruction input by a user, an instruction to analyze the query FIG resolution structure of the query instruction, the query instruction semantic content expansion, in order to better guide the search engine to select high-quality resources to meet the needs of the user's search, allowing users to search more efficient, saving network traffic.

附图说明 BRIEF DESCRIPTION

[0140] 图1是本发明搜索引擎与客户端实现互动的工作原理图; [0140] FIG. 1 is a search engine of the present invention interact with the client's operating principle;

[0141] 图2是本发明搜索引擎第一实施方式的模块图; [0141] FIG. 2 is a block diagram of the present invention, the search engine of the first embodiment;

[0142] 图3是本发明搜索引擎第二实施方式的模块图; [0142] FIG. 3 is a block diagram of the embodiment of the present invention the search engine a second embodiment;

[0143] 图4是本发明搜索引擎第三实施方式的模块图; [0143] FIG. 4 is a block diagram of the present invention, the search engine of the third embodiment;

[0144] 图5是本发明搜索引擎第四实施方式的模块图; [0144] FIG. 5 is a block diagram of the present invention, the search engine of the fourth embodiment;

[0145] 图6是本发明知识库架构的示意图; [0145] FIG. 6 is a schematic diagram of the architecture of the present invention, the knowledge base;

[0146] 图7是本发明搜索方法第一实施方式的流程图; [0146] FIG. 7 is a flowchart of a search method of a first embodiment of the present invention;

[0147] 图8是本发明搜索方法第二实施方式的流程图; [0147] FIG 8 is a flowchart of a search method of a second embodiment of the present invention;

[0148] 图9是本发明搜索方法第三实施方式的流程图; [0148] FIG. 9 is a flowchart of a search method of the third embodiment of the present invention;

[0149] 图10是本发明搜索方法第四实施方式的流程图; [0149] FIG. 10 is a flowchart of a search method of the present invention according to the fourth embodiment;

[0150] 图11是本发明“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”步骤一实施方式的流程图; [0150] FIG. 11 is the present invention, "Knowledge-based instruction needs the query intent analysis, specific needs of the intended query command" embodiment of a flowchart of the steps;

[0151] 图12是本发明“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”步骤另一实施方式的流程图; [0151] FIG. 12 is the present invention, "Knowledge-based instruction needs the query intent analysis, specific needs of the intended query command" step flow chart of another embodiment;

[0152] 图13是本发明表达模板库的构建方法的流程图; [0152] FIG. 13 is a flowchart of a method of constructing the present invention, the expression template library;

[0153] 图14是本发明在线界面一实施方式的示意图; [0153] FIG. 14 is a schematic diagram of a line interface of the present embodiment of the invention;

[0154] 图15是“当用户点击垂直搜索结果跳转至新页面”中的新页面示意图。 [0154] FIG. 15 is a "vertical search results when the user clicks a new page jump to" the new page in the diagram.

具体实施方式 detailed description

[0155] 以下将结合附图所示的各实施方式对本发明进行详细描述。 [0155] The present invention will hereinafter be described in detail in conjunction with the embodiments shown in the drawings. 但这些实施方式并不限制本发明,本领域的普通技术人员根据这些实施方式所轻易做出的结构、方法、或功能上的变换均包含在本发明的保护范围内。 However, these embodiments do not limit the present invention, the present structure made of ordinary skill in the art readily accordance with these embodiments, method, function, or conversion are included in the scope of the present invention.

[0156] 图1所示的本发明的搜索引擎10与客户端20实现互动的工作原理图。 [0156] FIG. 1 according to the present invention, the search engine 10 shown in FIG. 20 works to interact with the client. 本实施方式中,该客户端20包括一浏览器201,客户可通过该浏览器201打开搜索引擎在线展示的网页,并在网页中的对话框内输入查询指令,一般的,该输入的查询指令为文本信息,当然, 该查询指令还可以为图片信息、视频信息等等。 In the present embodiment, the client 20 includes a browser 201, the client 201 can display the online search engines to open the page by the browser and entering a query command within the page dialog general, the query command is input as a text message, of course, the query command may also be picture information, video information and so on. 所述搜索引擎10通过网络接收客户输入至所述浏览器中的查询指令,并对该查询指令进行搜索后,将搜索结果通过搜索引擎在线展示网页返回至该浏览器201。 After the engine 10 is input to the search query instruction by the browser receives the client network, and the search query instruction, the search engine will display the search results returned to the online web browser 201. 其中,该搜索引擎10可以包括一台或多台服务器,该客户端20可以包括一个或多个用户终端设备,如个人计算机、笔记本电脑、无线电话、个人数字处理(PDA)、或其它计算机系统和通信系统。 Wherein the search engine 10 may include one or more servers, other computer systems, the client 20 may include one or more user terminal devices, such as personal computers, laptop computers, wireless telephones, personal digital assistant (PDA), or and communication systems.

[0157] 这些服务器和终端设备在架构上都包含一些基本组件,如总线、处理系统、存储系统、一个或多个输入/输出系统、和通信接口等。 [0157] The server and the terminal device in the schema contains a number of basic components, such as a bus, a processing system, storage system, one or more input / output system, and a communication interface. 总线可以包括一个或多个导线,用来实现服务器或终端设备各组件之间的通信。 Bus may include one or more wires, used to implement communication between the server or the terminal device components. 处理系统包括各类型的用来执行指令、处理进程或线程的处理器或微处理器。 Processing system includes a processor or microprocessor for executing instructions of various types, the processing processes or threads. 存储系统可以包括存储动态信息的随机访问存储器(RAM)等动态存储器,和存储静态信息的只读存储器(ROM)等静态存储器,以及包括磁或光学记录介质与相应驱动的大容量存储器。 The system memory may include random access memory storing dynamic information (RAM), dynamic memory, etc., storing static information and read only memory (ROM) that static memory and the like, and comprises a magnetic or optical recording medium and the corresponding drive mass storage. 输入系统供用户输入信息到服务器或终端设备,如键盘、鼠标、手写笔、声音识别系统、或生物测定系统等。 An input system for user input information to a server or a terminal device, such as a keyboard, mouse, pen, voice recognition systems, or biometric systems. 输出系统包括用来输出信息的显示器、打印机、扬声器等。 Output system for outputting information includes a display, a printer, a speaker, and the like. 通信接口用来使服务器或终端设备与其它系统或系统进行通信。 A communication interface to allow a server or the terminal device to communicate with other systems or systems. 通信接口之间可通过有线连接、无线连接、或光连接连接到网络中,使搜索引擎10、客户端20间能够通过网络实现相互间的通信。 May be connected via a communication interface between a wired connection, wireless, or optical connection to the network, so that the search engine 10, enables the client 20 communicate with each other via a network. 网络可以包括局域网(LAN)、广域网(WAN)、电话网络如公共交换电话网(PSTN)、企业内部的互联网、因特网、或上述这些网络的结合等。 The network may include a local area network (LAN), wide area network (WAN), such as the Public Switched Telephone Network (PSTN), the Internet within the enterprise, the Internet, such as telephone network or combination of these networks.

[0158] 服务器和终端设备上均包含有用来管理系统资源、控制其它程序运行的操作系统软件,以及用来实现特定功能模块的应用软件。 [0158] both on the server and the terminal device comprises a system for managing resources, the control operating system software running of the program, and the software used to implement application-specific functional module. 如图2所示,在本发明第一实施方式中,所述搜索引擎包括了web服务模块101、与web服务模块101交互通信的UI模块102、与所述UI模块102通信的需求意图分析模块103、与所述需求意图分析模块103通信的结构分析模块104、与所述结构分析模块104通信的搜索模块105,以及与所述需求意图分析模块103、所述结构分析模块104交互通信的知识库106、与所述知识库106通信的用户历史行为库107、与所述知识库106、用户历史行为库107通信的表达模板挖掘模块108、与所述表达模板挖掘模块108和所述需求意图分析模块103通信的表达模板库109,以及与所述搜索模块105通信的网页存储库110。 2, in a first embodiment of the present invention, the search engine module 101 comprises a web service, the web interaction with the UI module 101 in communication service module 102, the UI module 102 and communications demand analysis module intent 103, the demand analysis module 104 analyzes the structure is intended to a communication module 103, the structure analysis module 104 searches the communication module 105, analysis module 103 and is intended to demand the knowledge of the interaction structure analysis module 104 to communicate library 106, the user behavior history database 106 in communication with the repository 107, and 106 the knowledge, the expression template mining module 107 communicating with the user behavior history database 108, and the expression of the template and excavation module 108 needs intent page repository 110 109, and the communication of the communication module 103 analyzes the search expression template library module 105. 值得一提的是,这些模块即可存储并运行于同一服务器中, 也可存储并运行在多台服务器中。 It is worth mentioning that these modules can be stored and run on the same server can also be stored and run on multiple servers.

[0159] 所述web服务模块101用于通过网络协议接收从客户端20传来的查询指令,并将该查询指令转到UI模块102,另外,该web服务模块101还用于接收所述UI模块102返回的结果页面,并将所述结果页面返回至客户端20。 [0159] The web service module 101 receives the query command 20 transmitted from a client through a network protocol, and the query instruction to the UI module 102, Further, the web service module 101 is further configured to receive the UI module 102 returns the result page, and the results page returned to the client 20.

[0160] 所述UI模块102用于接收所述web服务模块101传送的查询指令,并将该查询指令发送至所述查询指令分析模块103 ;另外,所述UI模块102还用于接收所述搜索模块104 返回的搜索结果,并将所述搜索结果拼装为结果页面后,返回所述结果页面至所述web服务模块101。 [0160] The UI module 102 for receiving the web service module 101 transmits a query instruction, and sends the inquiry command to the query command analysis module 103; Further, the UI module 102 is further configured to receive the search module 104 returns the search results, and the search results page after assembly as a result, returns the results page to the web service module 101.

[0161] 所述需求意图分析模块103用于调用所述知识库106、用户历史行为库107,以及所述表达模板库108,以对接收到的查询指令进行需求意图分析,明确所述查询指令的需求意图。 [0161] The intent demand analysis module 103 is used to call the Knowledge Base 106, user behavior history database 107, and the expression template library 108, to check the received instruction needs Intent Analysis, clear the query command It demands intentions. 在本发明中,所述意图分析模块103首先通过所述用户历史行为库107给所述知识库106中的各个知识片段的各个需求意图打分,具体的:用户在查询某类需求时,会点击相应的满足他需求的结果,如用户想获得汽车报价相关的信息,在搜索引擎输入查询指令“朗逸”后,会点击搜索引擎返回的汽车网站的tol,如“网易车库”,此时用户输入的查询指令“朗逸”片段以及用户点击的tol “网易车库”都隐含反映用户有找车相关信息的需求,基于这一点,本发明在计算各个知识片段的需求意图时,根据某个知识片段点击某类tol的数目/某个知识片段点击的总tol,来确定此知识片段需求意图的得分,如在所述用户历史行为库107中得知,查询指令为“朗逸”,其点击的总Url数为10个,其中,商品类tol为5 个,新闻类Url为3个,图片类Url为2个,则可计算出该查询指令 In the present invention, the analysis module 103 first intention by the user behavior history database to the individual requirements in each of the pieces of knowledge in the knowledge base 106 intention score 107, specifically: a user query certain requirements, will click corresponding to satisfy his demands results, such as the user wants to get car quotes related information, enter the query command in the search engine "Sunny" will click tol search engine returns automotive Web site, such as "Netease garage", then the user input query command "Sunny" fragment and tol user clicks the "NetEase garage" implicitly reflect user needs to find car-related information, based on this, the present invention in the calculation of the various pieces of knowledge of the intention of the demand, according to some pieces of knowledge tol certain number of clicks / click on a fragment of the total tol knowledge, this knowledge to determine the needs of the intention of scoring fragments, known as the user behavior history database 107, the query command is "Sunny", the click of the total Url number 10, wherein the commodity is tol 5, three news Url, Url picture type is 2, the query command can be calculated 商品类的需求意图为0. 5,新闻类的需求意图为0. 3,图片类的需求意图为0. 2 ;其次,在接收到用户输入的一个查询指令后,需要经过知识库106的知识匹配,得到所述查询指令中存在于所述知识库106 中的知识片段,并综合计算所述查询指令的知识库整体需求强度。 Commodity demand intention is 0.5, the demand for news is intent 0.3, demand is intent picture category 0.2; secondly, after receiving a query command entered by the user, need to go through knowledge repository 106 matching the query to obtain the strength of the overall demand instruction is present in the knowledge base knowledge base knowledge segment 106, and calculates the integrated query command. 例如用户输入“上海大众朗逸报价”,则通过知识库可获得“上海大众” “朗逸”的知识片段。 For example, the user inputs "Shanghai Volkswagen Lavida quote," through the knowledge base "Shanghai Volkswagen", "Sunny" fragments of knowledge available. 获得初步信息后,首先将知识片段“上海大众”和“朗逸”各自的需求意图得分加总,得到第一分数,其次,再通过知识片段“上海大众”与“朗逸”的关系,加减所述第一分数,得到知识库整体需求得分, 在本发明最佳实施方式中,若知识片段的关系为同属关系,则加分;若知识片段为非同属关系,则减分;若该知识库整体需求得分大于设定的阈值,则以知识库整体需求得分最高的需求类型作为查询指令的需求意图,并根据所述需求意图在所述查询指令中加入相应的tag 信息,例如“商品”、“新闻”、“图片”等。 After obtaining preliminary information, first of all the pieces of knowledge "Shanghai Volkswagen" and "Sunny" their needs intention score summed to obtain a first fraction, and secondly, through the pieces of knowledge Relations "Shanghai Volkswagen" and "Sunny", the subtraction of the said first fraction, to obtain the overall score needs knowledge, in the preferred embodiment of the present invention, if the relationship between knowledge fragments belong to the relationship, the points; if knowledge fragments belong to the non-relationship, it is less points; if the repository overall demand score is greater than a set threshold, places the highest scoring overall demand knowledge requirement types intended as query command, and the tag information added to the corresponding instruction in the query according to the intended requirements, e.g., "merchandise", "News", "picture" and so on. 值得一提的是:在本发明的最佳实施方式中,除了计算知识库整体需求得分外,在分析需求意图时,还会考虑表达模板层面上的打分:在接收到用户输入的一个查询指令后,需要经过表达模板库108的表达模板匹配,得到所述查询指令中存在于所述表达模板库108中的表达模板片段,例如用户输入“上海大众朗逸报价”, 则通过用户模板识别出查询指令中存在的“XX报价”模板。 It is worth mentioning: In a preferred embodiment of the present invention, in addition to the knowledge base to calculate the overall demand score, the intention in the analysis of demand, will also consider the level of expression on the template scoring: Upon receipt of a query entered by the user after the command, go through the expression template matching expression template database 108 to obtain the query instruction template fragments present in the template library 108 expression in the expression, for example, the user inputs "Sunny Shanghai Volkswagen Quote", the template identified by the user query command exists "XX offer" template. 在根据上述方法获得知识库整体需求得分的同时,查询指令又符合用户需求模板,表达模板库108也对所述查询指令在表达模板层面上进行打分,得到表达模板得分,则整体查询指令的需求强度得分为知识库整体需求得分与表达模板得分的加权之和,若该加权之和大于设定的阈值,则已加权之和最大的需求类型作为查询指令的需求意图,并根据所述需求意图在所述查询指令中加入相应的tag信息,例如“商品”、“新闻”、“图片”等。 While obtaining knowledge score overall demand by the above method, but also with user demand query instruction template, the template library 108 also expressing the query instructions on the template scoring expression level, resulting in the expression template scores, the overall query requirement command Knowledge intensity score of the score and the expression template overall demand weighted sum of the score, and if the weighted sum is greater than the set threshold, and the maximum weighted sum already needs demand type intended as query command, according to the needs and intentions adding the appropriate tag information in the query instructions, such as "commodity", "News", "picture" and so on.

[0162] 所述结构分类模块104用于结合所述知识库106,对经过所述需求意图分析模块103后的查询指令进行智能化变换后发送至所述搜索模块105,其中,所述智能化变换即是语义内容扩充,所述的语义内容扩充包括了语义内容的归一化,以及语义类别的扩展。 [0162] The classification module 104 is used in conjunction with the structure of the knowledge base 106, to the search module 105 transmits after conversion intelligently query instruction through the demand analysis module 103 is intended, wherein said intelligent semantic content conversion that is expanded, said expansion comprises the semantic content normalized, and the expansion of the semantic content of the semantic class. 具体的,在所述查询指令有同属关系(上位属性的知识片段+下位属性的知识片段)时,例如, 所述查询指令为“手机诺基亚”,此时,所述结构分类模块104在发送给所述搜索模块105 时,即会在“手机”这个上位属性的知识片段上加入“可以丢弃”的tag,这样,在搜索模块105对所述查询指令进行搜索时,即可通过“手机诺基亚”进行搜索,也可通过“诺基亚”进行搜索,且还可认为带有“手机诺基亚”文本信息的网页和只带有“诺基亚”文本信息的网页权值一样;另外,例如:若所述查询指令为“诺基亚”,则所述结构分类模块104在发送给所述搜索模块105时,还可根据“诺基亚”对应其上位属性进行扩展,如扩展为“手机”,这样,在搜索模块105对“诺基亚”进行搜索时,可根据结果的数量判断是否需要扩展到“手机”进行搜索,如通过“诺基亚 Specifically, when the query instructions belong have the relationship (+ property knowledge fragment upper lower segment attribute knowledge), for example, the query command is "Nokia" In this case, the structure 104 to a classification module the search module 105, which will be added in the knowledge fragment "mobile phone" in this upper property "may discard" tag, you so that, when the search module 105 searches the query instruction, to the "Nokia" search, can also be searched by "Nokia" and may also be considered with the "Nokia" web page and only with the "Nokia" the same page weights text information text information; in addition, for example: if the query command the "Nokia", then the structure of the classification module 104 to transmit the search module 105, may be expanded according to "Nokia" attributes corresponding to its upper position, such as extended as "mobile phone", so that, in the search module 105 on the " Nokia "when you search, the results may determine whether the number needs to be extended to the" mobile phone "searches, such as through the" Nokia 搜索到的结果数量较小时,即可扩展到“手机”;又如:若所述查询指令为“手机”,则所述结构分类模块104可将“手机”对应其同位属性进行扩展,如扩展为“电脑”,这样的扩展可用于广告的推广,如在搜索页面的右侧即可根据“手机”这个查询指令进行广告推广,又可根据“电脑”这个查询指令进行广告推广;再如,若所述查询指令为“诺基亚”,则所述结构分类模块104在发送给所述搜索模块105时,还可根据“诺基亚” 对应其下位属性进行扩展,如扩展为“N71”、“N72”等,这样,在搜索模块105在搜索到带有“Ν71”、“Ν72”等文本信息网页时,也可根据这些网页的权值,判断是否作为搜索结果输出。 When a small number of search results can be extended to the "phone"; Another example: if the query instruction is a "mobile phone", then the configuration module 104 may be classified "mobile phone" which corresponds to the same bit extended attributes, such as extended the "computer", such extensions can be used to promote advertising, such as advertising can be carried out according to the "mobile phone" in the search query instruction to the right of the page, but also for advertising based on "computer" this query command; Again, If the query command is "Nokia", then the structure of the classification module 104 is transmitted to the search module 105, may be expanded according to "Nokia" which corresponds to the lower properties, such as extended as "N71", "N72" etc. Thus, the search module 105 when searching the page with text information "Ν71", "Ν72", etc., but also according to the weight of these pages, determines whether the output as a search result. 这种权值判断可参现有搜索引擎中的权值判断,在此不再赘述。 Such weight determination reference value may be determined in conventional search engines weights, are not repeated here. 总而言之,对所述查询指令进行扩展时,可根据搜索的策略,即可扩展其上位属性的知识片段,也可扩展其同位属性的知识片段,也可扩展其下位属性的知识片段。 In summary, when the extended query command, according to a search strategy, which can extend the upper property knowledge fragment, which can also be extended property knowledge parity fragments, fragments can also be extended their knowledge of the properties lower.

[0163] 所述搜索模块105用于接收经过所述需求意图分析模块103或所述结构分类模块104的智能化变换后的查询指令,并将该查询指令在网页存储库110中搜索,以得到搜索结果,同时,所述搜索模块105还用于将所述搜索结果返回至所述UI模块102。 [0163] The search module 105 for receiving the structural classification module 103 or the needs of the intended analysis module 104 queries the intelligent conversion instruction, and the instruction a search query page repository 110, to give search result, while the search module 105 is further configured to return the search result to the UI module 102.

[0164] 所述知识库106用于存储先验知识。 [0164] The knowledge base 106 for storing a priori knowledge. 在本发明最佳实施方式中,所述知识库主要存储为树形结构,对每一类知识库构建一棵知识库树,通过该知识库树的父亲节点标识其上位属性,右兄弟节点表示其同位属性,左兄弟节点表示其下位属性,如此迭代,直至叶子节点。 In the preferred embodiment of the invention, the main memory tree structure knowledge, to build a knowledge base for each category tree knowledge base, identify its parent node through the upper property knowledge tree, right sibling node represents its parity property, said its sibling left lower property, so iterations, until a leaf node. 如图6所示,“大众”为其最上位属性;其下位属性为“上海大众”;与所述“上海大众” 同位的有“一汽大众”;在所述“上海大众”下位的有“朗逸”,与所述“朗逸”同位的有“途 "Mass" to its uppermost property shown in FIG. 6; the lower attribute is "Shanghai Volkswagen"; a "FAW public" and the "Shanghai Volkswagen" with the bit; have the "Shanghai Volkswagen" the lower " Sunny ", and the" Sunny "bits have the same" way

观”......这种知识库的构建方法,本领域的普通技术人员可参现有技术完成,在此不再赘述。 View "...... method of constructing such a knowledge base, those of ordinary skill in the prior art reference may be complete, are not repeated here.

[0165] 用户历史行为库107用于存储用户历史搜索记录。 [0165] user behavior history database 107 is used to store user search history records. 优选地,其可包括查询指令、查询次数,以及加权点击数等信息。 Preferably, it may include information query instruction, queries, and the weighting hits like.

[0166] 表达模板挖掘模块108用于根据所述知识库106中的知识片段和所述用户历史行为库107中的用户历史查询指令,挖掘出表达模板,并将所述表达模板存储于所述表达模板库109中。 [0166] Expression template mining module 108 according to the pieces of knowledge in the knowledge base 106 and the user behavior history of the user history database 107 query instruction, excavated expression template, and the template stored in the expression expression template library 109. 相同需求的用户,在表达方式上会出现相似性,所述表达模板是指,一般用户在有一定查询需求时,其输入的查询指令为何,例如,当用户在查询汽车相关信息时,表达方式会有:“速腾怎么样”、“马六动力如何”等,其中即可抽取出“【汽车品牌/型号】怎么样”、“【汽车品牌/型号】动力如何”等表达汽车需求时常用的表达模板。 The same user demand, there will be a similarity in the expression, the expression refers to a template, generally when the user needs a certain query, query instruction input why, for example, when the user information query automobile, expression there will be: "Sagitar how", "how to power six horses," and so on, which can be extracted "[car make / model] how", "[car make / model] how to power" and other commonly used expression when demand for cars expression template. 在本发明的一实施方式中,具体为:首先在所述用户历史行为库107中包含知识库106知识片段的查询指令抽取出来,如在“马六如何”、“斯柯达如何”、“速腾如何”的查询指令中,抽取出知识片段:“马六”、“斯柯达”、“速腾”,其次将知识库片段替换成“【汽车品牌/型号】”符号,即生成“【汽车品牌/型号】如何”的候选表达模板;再次,统计生成的候选表达模板符合的知识库片段的数量,若该数量大于设定的阈值,则将所述候选表达模板作为表达模板,存于所述表达模板库109中;若该数量小于设定的阈值,则舍弃所述候选表达模板。 In one embodiment of the present invention, specifically as follows: First, a query command comprises 106 pieces of knowledge in the knowledge base in the user behavior history database 107 extracted, as described in "How to six horses", "how Skoda", "How suteng "query command, extract the pieces of knowledge:" six horses "," Skoda "," Sagitar ", followed by the replacement knowledge fragment as" [car make / model] "symbol, which generates" [car make / model] how "expression candidate template; again, the number of segments of the knowledge base to generate statistical template matching expression candidate, if the number is greater than a set threshold, the template will be expressed as an expression of the candidate template stored in the template library expression 109; if the number is less than a set threshold, the candidate is discarded expression template.

[0167] 所述表达模板库109用于存储由所述表达模板挖掘模块108挖掘出的表达模板。 [0167] The expression template in the gallery 109 for storing the template expression excavated mining module 108.

[0168] 所述网页存储库110用于存储网页数据和该网页数据的索引信息。 [0168] The web page store 110 for storing index information of the web page data and page data. 该数据库即是普通搜索引擎常用的数据库,在此不再赘述。 The database that is commonly used in ordinary search engine database, which will not be repeated here.

[0169] 如图3所示,在本发明第二实施方式中,所述搜索引擎包括了web服务模块101、 与web服务模块101交互通信的UI模块102、与所述UI模块102通信的需求意图分析模块103、与所述需求意图分析模块103通信的结构分析模块104、与所述结构分析模块104通信的搜索模块105,以及与所述需求意图分析模块103、所述结构分析模块104交互通信的知识库106、与所述知识库106通信的用户历史行为库107、与所述知识库106、用户历史行为库107通信的表达模板挖掘模块108、与所述表达模板挖掘模块108和所述需求意图分析模块103通信的表达模板库109,以及与所述搜索模块105通信的网页存储库110、第一垂直 [0169] 3, in the second embodiment of the present invention, the search engine include web services module 101, the UI module 102 and the communication needs of the UI module and the web service communication module 101 to interact 102 intended analysis module 103, in communication with the analysis module 103 and is intended to demand structural analysis module 104, the search module 104 in communication structure analysis module 105, and the requirements and intended analysis module 103, analysis module 104 interaction with the structure knowledge base 106 in communication with the user behavior history database 106 in communication with the repository 107, and 106 the knowledge, the expression template mining module 107 communicating with the user behavior history database 108, and the expression of the template and excavation module 108 communicating said analysis module 103 needs intention expression template database 109, and a communication module 105 and the search web page repository 110, a first vertical

搜索数据111a、第二垂直搜索数据库Illb........第N垂直搜索数据库llln。 Search data 111a, a second vertical search database Illb ........ N-th vertical search database llln. 值得一提 Worth to talk about

的是,这些模块即可存储并运行于同一服务器中,也可存储并运行在多台服务器中。 That these modules can be stored and run on the same server can also be stored and run on multiple servers.

[0170] 所述web服务模块101用于通过网络协议接收从客户端20传来的查询指令,并将该查询指令转到UI模块102,另外,该web服务模块101还用于接收所述UI模块102返回的结果页面,并将所述结果页面返回至客户端20。 [0170] The web service module 101 receives the query command 20 transmitted from a client through a network protocol, and the query instruction to the UI module 102, Further, the web service module 101 is further configured to receive the UI module 102 returns the result page, and the results page returned to the client 20.

[0171] 所述UI模块102用于接收所述web服务模块101传送的查询指令,并将该查询指令发送至所述查询指令分析模块103 ;另外,所述UI模块102还用于接收所述搜索模块104 返回的搜索结果,并将所述搜索结果拼装为结果页面后,返回所述结果页面至所述web服务模块101。 [0171] The UI module 102 for receiving the web service module 101 transmits a query instruction, and sends the inquiry command to the query command analysis module 103; Further, the UI module 102 is further configured to receive the search module 104 returns the search results, and the search results page after assembly as a result, returns the results page to the web service module 101.

[0172] 所述需求意图分析模块103用于调用所述知识库106、用户历史行为库107,以及所述表达模板库108,以对接收到的查询指令进行需求意图分析,明确所述查询指令的需求意图。 [0172] The intent demand analysis module 103 is used to call the Knowledge Base 106, user behavior history database 107, and the expression template library 108, to check the received instruction needs Intent Analysis, clear the query command It demands intentions. 在本发明中,所述意图分析模块103首先通过所述用户历史行为库107给所述知识库106中的各个知识片段的各个需求意图打分,具体的:用户在查询某类需求时,会点击相应的满足他需求的结果,如用户想获得汽车报价相关的信息,在搜索引擎输入查询指令“朗逸”后,会点击搜索引擎返回的汽车网站的tol,如“网易车库”,此时用户输入的查询指令“朗逸”片段以及用户点击的tol “网易车库”都隐含反映用户有找车相关信息的需求,基于这一点,本发明在计算各个知识片段的需求意图时,根据某个知识片段点击某类tol的数目/某个知识片段点击的总此1,来确定此知识片段需求意图的得分,如在所述用户历史行为库107中得知,查询指令为“朗逸”,其点击的总Url数为10个,其中,商品类tol为5 个,新闻类Url为3个,图片类Url为2个,则可计算出该查询指令 In the present invention, the analysis module 103 first intention by the user behavior history database to the individual requirements in each of the pieces of knowledge in the knowledge base 106 intention score 107, specifically: a user query certain requirements, will click corresponding to satisfy his demands results, such as the user wants to get car quotes related information, enter the query command in the search engine "Sunny" will click tol search engine returns automotive Web site, such as "Netease garage", then the user input query command "Sunny" fragment and tol user clicks the "NetEase garage" implicitly reflect user needs to find car-related information, based on this, the present invention in the calculation of the various pieces of knowledge of the intention of the demand, according to some pieces of knowledge tol certain number of clicks / total 1 a knowledge of this segment clicks to determine this segment demand the knowledge of the intention of scoring, as known in the user behavior history database 107, query instruction is "Sunny", with its click Url total of 10, wherein the commodity is tol 5, three news Url, Url picture type is 2, the query command can be calculated 商品类的需求意图为0. 5,新闻类的需求意图为0. 3,图片类的需求意图为0. 2 ;其次,在接收到用户输入的一个查询指令后,需要经过知识库106的知识匹配,得到所述查询指令中存在于所述知识库106 中的知识片段,并综合计算所述查询指令的知识库整体需求强度。 Commodity demand intention is 0.5, the demand for news is intent 0.3, demand is intent picture category 0.2; secondly, after receiving a query command entered by the user, need to go through knowledge repository 106 matching the query to obtain the strength of the overall demand instruction is present in the knowledge base knowledge base knowledge segment 106, and calculates the integrated query command. 例如用户输入“上海大众朗逸报价”,则通过知识库可获得“上海大众” “朗逸”的知识片段。 For example, the user inputs "Shanghai Volkswagen Lavida quote," through the knowledge base "Shanghai Volkswagen", "Sunny" fragments of knowledge available. 获得初步信息后,首先将知识片段“上海大众”和“朗逸”各自的需求意图得分加总,得到第一分数,其次,再通过知识片段“上海大众”与“朗逸”的关系,加减所述第一分数,得到知识库整体需求得分, 在本发明最佳实施方式中,若知识片段的关系为同属关系,则加分;若知识片段为非同属关系,则减分;若该知识库整体需求得分大于设定的阈值,则以知识库整体需求得分最高的需求类型作为查询指令的需求意图,并根据所述需求意图在所述查询指令中加入相应的tag 信息,例如“商品”、“新闻”、“图片”等。 After obtaining preliminary information, first of all the pieces of knowledge "Shanghai Volkswagen" and "Sunny" their needs intention score summed to obtain a first fraction, and secondly, through the pieces of knowledge Relations "Shanghai Volkswagen" and "Sunny", the subtraction of the said first fraction, to obtain the overall score needs knowledge, in the preferred embodiment of the present invention, if the relationship between knowledge fragments belong to the relationship, the points; if knowledge fragments belong to the non-relationship, it is less points; if the repository overall demand score is greater than a set threshold, places the highest scoring overall demand knowledge requirement types intended as query command, and the tag information added to the corresponding instruction in the query according to the intended requirements, e.g., "merchandise", "News", "picture" and so on. 值得一提的是:在本发明的最佳实施方式中,除了计算知识库整体需求得分外,在分析需求意图时,还会考虑表达模板层面上的打分:在接收到用户输入的一个查询指令后,需要经过表达模板库108的表达模板匹配,得到所述查询指令中存在于所述表达模板库108中的表达模板片段,例如用户输入“上海大众朗逸报价”, 则通过用户模板识别出查询指令中存在的“XX报价”模板。 It is worth mentioning: In a preferred embodiment of the present invention, in addition to the knowledge base to calculate the overall demand score, the intention in the analysis of demand, will also consider the level of expression on the template scoring: Upon receipt of a query entered by the user after the command, go through the expression template matching expression template database 108 to obtain the query instruction template fragments present in the template library 108 expression in the expression, for example, the user inputs "Sunny Shanghai Volkswagen Quote", the template identified by the user query command exists "XX offer" template. 在根据上述方法获得知识库整体需求得分的同时,查询指令又符合用户需求模板,表达模板库108也对所述查询指令在表达模板层面上进行打分,得到表达模板得分,则整体查询指令的需求强度得分为知识库整体需求得分与表达模板得分的加权之和,若该加权之和大于设定的阈值,则已加权之和最大的需求类型作为查询指令的需求意图。 While obtaining knowledge score overall demand by the above method, but also with user demand query instruction template, the template library 108 also expressing the query instructions on the template scoring expression level, resulting in the expression template scores, the overall query requirement command strength score score score expression template weighted sum of overall demand knowledge, if the weighted sum is greater than the set threshold, and the maximum weighted sum already needs demand type intended as query command.

[0173] 所述结构分类模块104用于结合所述知识库106,对经过所述需求意图分析模块103后的查询指令进行智能化变换后发送至所述搜索模块105,其中,所述智能化变换即是语义内容扩充,所述的语义内容扩充包括了语义内容的归一化,以及语义类别的扩展。 [0173] The classification module 104 is used in conjunction with the structure of the knowledge base 106, to the search module 105 transmits after conversion intelligently query instruction through the demand analysis module 103 is intended, wherein said intelligent semantic content conversion that is expanded, said expansion comprises the semantic content normalized, and the expansion of the semantic content of the semantic class. 具体的,在所述查询指令有同属关系(上位属性的知识片段+下位属性的知识片段)时,例如, 所述查询指令为“手机诺基亚”,此时,所述结构分类模块104在发送给所述搜索模块105 时,即会在“手机”这个上位属性的知识片段上加入“可以丢弃”的tag,这样,在搜索模块105对所述查询指令进行搜索时,即可通过“手机诺基亚”进行搜索,也可通过“诺基亚”进行搜索,且还可认为带有“手机诺基亚”文本信息的网页和只带有“诺基亚”文本信息的网页权值一样;另外,例如:若所述查询指令为“诺基亚”,则所述结构分类模块104在发送给所述搜索模块105时,还可根据“诺基亚”对应其上位属性进行扩展,如扩展为“手机”,这样,在搜索模块105对“诺基亚”进行搜索时,可根据结果的数量判断是否需要扩展到“手机”进行搜索,如通过“诺基亚 Specifically, when the query instructions belong have the relationship (+ property knowledge fragment upper lower segment attribute knowledge), for example, the query command is "Nokia" In this case, the structure 104 to a classification module the search module 105, which will be added in the knowledge fragment "mobile phone" in this upper property "may discard" tag, you so that, when the search module 105 searches the query instruction, to the "Nokia" search, can also be searched by "Nokia" and may also be considered with the "Nokia" web page and only with the "Nokia" the same page weights text information text information; in addition, for example: if the query command the "Nokia", then the structure of the classification module 104 to transmit the search module 105, may be expanded according to "Nokia" attributes corresponding to its upper position, such as extended as "mobile phone", so that, in the search module 105 on the " Nokia "when you search, the results may determine whether the number needs to be extended to the" mobile phone "searches, such as through the" Nokia 搜索到的结果数量较小时,即可扩展到“手机”;又如:若所述查询指令为“手机”,则所述结构分类模块104可将“手机”对应其同位属性进行扩展,如扩展为“电脑”,这样的扩展可用于广告的推广,如在搜索页面的右侧即可根据“手机”这个查询指令进行广告推广,又可根据“电脑”这个查询指令进行广告推广;再如,若所述查询指令为“诺基亚”,则所述结构分类模块104在发送给所述搜索模块105时,还可根据“诺基亚” 对应其下位属性进行扩展,如扩展为“N71”、“N72”等,这样,在搜索模块105在搜索到带有“Ν71”、“Ν72”等文本信息网页时,也可根据这些网页的权值,判断是否作为搜索结果输出。 When a small number of search results can be extended to the "phone"; Another example: if the query instruction is a "mobile phone", then the configuration module 104 may be classified "mobile phone" which corresponds to the same bit extended attributes, such as extended the "computer", such extensions can be used to promote advertising, such as advertising can be carried out according to the "mobile phone" in the search query instruction to the right of the page, but also for advertising based on "computer" this query command; Again, If the query command is "Nokia", then the structure of the classification module 104 is transmitted to the search module 105, may be expanded according to "Nokia" which corresponds to the lower properties, such as extended as "N71", "N72" etc. Thus, the search module 105 when searching the page with text information "Ν71", "Ν72", etc., but also according to the weight of these pages, determines whether the output as a search result. 这种权值判断可参现有搜索引擎中的权值判断,在此不再赘述。 Such weight determination reference value may be determined in conventional search engines weights, are not repeated here. 总而言之,对所述查询指令进行扩展时,可根据搜索的策略,即可扩展其上位属性的知识片段,也可扩展其同位属性的知识片段,也可扩展其下位属性的知识片段。 In summary, when the extended query command, according to a search strategy, which can extend the upper property knowledge fragment, which can also be extended property knowledge parity fragments, fragments can also be extended their knowledge of the properties lower.

[0174] 所述搜索模块105用于接收经过所述需求意图分析模块103或所述结构分类模块104的智能化变换后的查询指令,并将该查询指令在多个垂直搜索数据库(第一垂直搜索 [0174] The search module 105 for receiving the demands of the intended structure analysis module 103 or the classification module 104 queries the intelligent conversion instruction, and the instruction in the plurality of vertical search query databases (first vertical search for

数据库111a、第二垂直搜索数据库Illb........第N垂直数据库llln)的其中之一,以及 Database 111a, a second vertical search database Illb ........ N-th vertical database llln) wherein one, and

所述网页存储库1110中搜索,以得到搜索结果,同时,所述搜索模块105还用于将所述搜索结果返回至所述UI模块102。 The web page repository 1110 searches, to obtain search results, and the search module 105 is further configured to return the search result to the UI module 102. 值得一提的是:选择某个垂直搜索数据库进行垂直搜索是通过查询指令的需求意图确定的,例如:若查询指令的需求意图为“商品”,则在商品垂直搜索数据库中进行搜索;所查询指令的需求意图为“图片”,则在图片垂直搜索数据库中进行搜索,其中,在垂直搜索数据库中搜索到的一条或多条结果,会插入至在网页存储库中搜索到的结果中,形成整体搜索结果。 It is worth mentioning: select a vertical search vertical search databases by querying the command is intent demand determined, for example: If the query command is intended to demand "commodity", the search vertical search in the commodity database; the query demand intended instructions is "image", the search is performed in the vertical image search database, wherein, in the searched vertical search results database and one or more, is inserted into the page to the search result repository, the formation of the overall search results. 所述垂直搜索,即是在某个特定的类别下进行搜索,其具体的搜索方法和系统在本领域中已多有现有技术揭示,在此不再赘述。 The vertical search that is carried out in the search for a specific category, the specific search method and system of the present art has revealed more than a prior art, it is not repeated here.

[0175] 所述知识库106用于存储先验知识。 [0175] The knowledge base 106 for storing a priori knowledge. 在本发明最佳实施方式中,所述知识库主要存储为树形结构,对每一类知识库构建一棵知识库树,通过该知识库树的父亲节点标识其上位属性,右兄弟节点表示其同位属性,左兄弟节点表示其下位属性,如此迭代,直至叶子节点。 In the preferred embodiment of the invention, the main memory tree structure knowledge, to build a knowledge base for each category tree knowledge base, identify its parent node through the upper property knowledge tree, right sibling node represents its parity property, said its sibling left lower property, so iterations, until a leaf node. 如图6所示,“大众”为其最上位属性;其下位属性为“上海大众”;与所述“上海大众” 同位的有“一汽大众”;在所述“上海大众”下位的有“朗逸”,与所述“朗逸”同位的有“途 "Mass" to its uppermost property shown in FIG. 6; the lower attribute is "Shanghai Volkswagen"; a "FAW public" and the "Shanghai Volkswagen" with the bit; have the "Shanghai Volkswagen" the lower " Sunny ", and the" Sunny "bits have the same" way

观”......这种知识库的构建方法,本领域的普通技术人员可参现有技术完成,在此不再赘述。[0176] 用户历史行为库107用于存储用户历史搜索记录。优选地,其可包括查询指令、查询次数,以及加权点击数等信息。 View "...... method of constructing such a knowledge base, those of ordinary skill in the prior art reference may be complete, are not repeated here. [0176] user behavior history database 107 for storing user history search history preferably, it may include information query instructions, queries, and the weighting hits like.

[0177] 表达模板挖掘模块108用于根据所述知识库106中的知识片段和所述用户历史行为库107中的用户历史查询指令,挖掘出表达模板,并将所述表达模板存储于所述表达模板库109中。 [0177] Expression template mining module 108 according to the pieces of knowledge in the knowledge base 106 and the user behavior history of the user history database 107 query instruction, excavated expression template, and the template stored in the expression expression template library 109. 相同需求的用户,在表达方式上会出现相似性,所述表达模板是指,一般用户在有一定查询需求时,其输入的查询指令为何,例如,当用户在查询汽车相关信息时,表达方式会有:“速腾怎么样”、“马六动力如何”等,其中即可抽取出“【汽车品牌/型号】怎么样”、“【汽车品牌/型号】动力如何”等表达汽车需求时常用的表达模板。 The same user demand, there will be a similarity in the expression, the expression refers to a template, generally when the user needs a certain query, query instruction input why, for example, when the user information query automobile, expression there will be: "Sagitar how", "how to power six horses," and so on, which can be extracted "[car make / model] how", "[car make / model] how to power" and other commonly used expression when demand for cars expression template. 在本发明的一实施方式中,具体为:首先在所述用户历史行为库107中包含知识库106知识片段的查询指令抽取出来,如在“马六如何”、“斯柯达如何”、“速腾如何”的查询指令中,抽取出知识片段:“马六”、“斯柯达”、“速腾”,其次将知识库片段替换成“【汽车品牌/型号】”符号,即生成“【汽车品牌/型号】如何”的候选表达模板;再次,统计生成的候选表达模板符合的知识库片段的数量,若该数量大于设定的阈值,则将所述候选表达模板作为表达模板,存于所述表达模板库109中;若该数量小于设定的阈值,则舍弃所述候选表达模板。 In one embodiment of the present invention, specifically as follows: First, a query command comprises 106 pieces of knowledge in the knowledge base in the user behavior history database 107 extracted, as described in "How to six horses", "how Skoda", "How suteng "query command, extract the pieces of knowledge:" six horses "," Skoda "," Sagitar ", followed by the replacement knowledge fragment as" [car make / model] "symbol, which generates" [car make / model] how "expression candidate template; again, the number of segments of the knowledge base to generate statistical template matching expression candidate, if the number is greater than a set threshold, the template will be expressed as an expression of the candidate template stored in the template library expression 109; if the number is less than a set threshold, the candidate is discarded expression template.

[0178] 所述表达模板库109用于存储由所述表达模板挖掘模块108挖掘出的表达模板。 [0178] The expression template in the gallery 109 for storing the template expression excavated mining module 108.

[0179] 所述网页存储库110用于存储网页数据和该网页数据的索引信息。 [0179] The web page store 110 for storing index information of the web page data and page data. 该数据库即是普通搜索引擎常用的数据库,在此不再赘述。 The database that is commonly used in ordinary search engine database, which will not be repeated here.

[0180] 所述第一垂直搜索数据库11 la、第二垂直搜索数据库Illb........第N垂直搜索 [0180] The first vertical search database 11 la, the second vertical search database Illb ........ vertical search of N

数据库Illn用于存储特定类别数据和该特定类别数据的索引信息,例如商品数据、商品索引;新闻数据、新闻索引;图片数据、图片索引等。 Illn index database for storing information about specific categories of data and the specific categories of data, such as product data, commodity index; news, news index; picture data, images and other indexes.

[0181] 如图4所示,在本发明第三实施方式中,所述搜索引擎包括了web服务模块101、 与web服务模块101交互通信的UI模块102、与所述UI模块102通信的需求意图分析模块103、与所述UI模块102通信的结构分析模块104、与所述结构分析模块104通信的搜索模块105,以及与所述需求意图分析模块103、所述结构分析模块104交互通信的知识库106、 与所述知识库106通信的用户历史行为库107、与所述知识库106、用户历史行为库107通信的表达模板挖掘模块108、与所述表达模板挖掘模块108和所述需求意图分析模块103通信的表达模板库109,以及与所述搜索模块105通信的网页存储库110。 [0181] As shown, in the third embodiment of the present invention, the search engine 4 includes web service module 101, the UI module 102 and the communication needs of web interaction with the UI module 101 in communication service module 102 intended analysis module 103, the UI module 102 and the communication structure analysis module 104, the search module 104 in communication structure analysis module 105, and the requirements and intended analysis module 103, the interactive communication structure analysis module 104 knowledge base 106, and the user behavior history database 106 in communication with the repository 107, and 106 the knowledge, the expression template mining module 107 communicating with the user behavior history database 108, and the expression of the template and excavation module 108 needs page repository 110 109, and the communication of the communication module 103 is intended to analyze the expression template database 105 to the search module. 值得一提的是,这些模块即可存储并运行于同一服务器中,也可存储并运行在多台服务器中。 It is worth mentioning that these modules can be stored and run on the same server can also be stored and run on multiple servers.

[0182] 所述web服务模块101用于通过网络协议接收从客户端20传来的查询指令,并将该查询指令转到UI模块102,另外,该web服务模块101还用于接收所述UI模块102返回的结果页面,并将所述结果页面返回至客户端20。 [0182] The web service module 101 receives the query command 20 transmitted from a client through a network protocol, and the query instruction to the UI module 102, Further, the web service module 101 is further configured to receive the UI module 102 returns the result page, and the results page returned to the client 20.

[0183] 所述UI模块102用于接收所述web服务模块101传送的查询指令,并将该查询指令发送至所述查询指令分析模块103 ;另外,所述UI模块102还用于接收所述搜索模块104 返回的搜索结果,并将所述搜索结果拼装为结果页面后,返回所述结果页面至所述web服务模块101。 [0183] The UI module 102 for receiving the web service module 101 transmits a query instruction, and sends the inquiry command to the query command analysis module 103; Further, the UI module 102 is further configured to receive the search module 104 returns the search results, and the search results page after assembly as a result, returns the results page to the web service module 101.

[0184] 所述需求意图分析模块103用于调用所述知识库106、用户历史行为库107,以及所述表达模板库108,以对接收到的查询指令进行需求意图分析,明确所述查询指令的需求意图。 [0184] The intent demand analysis module 103 is used to call the Knowledge Base 106, user behavior history database 107, and the expression template library 108, to check the received instruction needs Intent Analysis, clear the query command It demands intentions. 在本发明中,所述意图分析模块103首先通过所述用户历史行为库107给所述知识库106中的各个知识片段的各个需求意图打分,具体的:用户在查询某类需求时,会点击相应的满足他需求的结果,如用户想获得汽车报价相关的信息,在搜索引擎输入查询指令“朗逸”后,会点击搜索引擎返回的汽车网站的tol,如“网易车库”,此时用户输入的查询指令“朗逸”片段以及用户点击的tol “网易车库”都隐含反映用户有找车相关信息的需求,基于这一点,本发明在计算各个知识片段的需求意图时,根据某个知识片段点击某类tol的数目/某个知识片段点击的总此1,来确定此知识片段需求意图的得分,如在所述用户历史行为库107中得知,查询指令为“朗逸”,其点击的总Url数为10个,其中,商品类tol为5 个,新闻类Url为3个,图片类Url为2个,则可计算出该查询指令 In the present invention, the analysis module 103 first intention by the user behavior history database to the individual requirements in each of the pieces of knowledge in the knowledge base 106 intention score 107, specifically: a user query certain requirements, will click corresponding to satisfy his demands results, such as the user wants to get car quotes related information, enter the query command in the search engine "Sunny" will click tol search engine returns automotive Web site, such as "Netease garage", then the user input query command "Sunny" fragment and tol user clicks the "NetEase garage" implicitly reflect user needs to find car-related information, based on this, the present invention in the calculation of the various pieces of knowledge of the intention of the demand, according to some pieces of knowledge tol certain number of clicks / total 1 a knowledge of this segment clicks to determine this segment demand the knowledge of the intention of scoring, as known in the user behavior history database 107, query instruction is "Sunny", with its click Url total of 10, wherein the commodity is tol 5, three news Url, Url picture type is 2, the query command can be calculated 商品类的需求意图为0. 5,新闻类的需求意图为0. 3,图片类的需求意图为0. 2 ;其次,在接收到用户输入的一个查询指令后,需要经过知识库106的知识匹配,得到所述查询指令中存在于所述知识库106 中的知识片段,并综合计算所述查询指令的知识库整体需求强度。 Commodity demand intention is 0.5, the demand for news is intent 0.3, demand is intent picture category 0.2; secondly, after receiving a query command entered by the user, need to go through knowledge repository 106 matching the query to obtain the strength of the overall demand instruction is present in the knowledge base knowledge base knowledge segment 106, and calculates the integrated query command. 例如用户输入“上海大众朗逸报价”,则通过知识库可获得“上海大众” “朗逸”的知识片段。 For example, the user inputs "Shanghai Volkswagen Lavida quote," through the knowledge base "Shanghai Volkswagen", "Sunny" fragments of knowledge available. 获得初步信息后,首先将知识片段“上海大众”和“朗逸”各自的需求意图得分加总,得到第一分数,其次,再通过知识片段“上海大众”与“朗逸”的关系,加减所述第一分数,得到知识库整体需求得分, 在本发明最佳实施方式中,若知识片段的关系为同属关系,则加分;若知识片段为非同属关系,则减分;若该知识库整体需求得分大于设定的阈值,则以知识库整体需求得分最高的需求类型作为查询指令的需求意图,并根据所述需求意图在所述查询指令中加入相应的tag 信息,例如“商品”、“新闻”、“图片”等。 After obtaining preliminary information, first of all the pieces of knowledge "Shanghai Volkswagen" and "Sunny" their needs intention score summed to obtain a first fraction, and secondly, through the pieces of knowledge Relations "Shanghai Volkswagen" and "Sunny", the subtraction of the said first fraction, to obtain the overall score needs knowledge, in the preferred embodiment of the present invention, if the relationship between knowledge fragments belong to the relationship, the points; if knowledge fragments belong to the non-relationship, it is less points; if the repository overall demand score is greater than a set threshold, places the highest scoring overall demand knowledge requirement types intended as query command, and the tag information added to the corresponding instruction in the query according to the intended requirements, e.g., "merchandise", "News", "picture" and so on. 值得一提的是:在本发明的最佳实施方式中,除了计算知识库整体需求得分外,在分析需求意图时,还会考虑表达模板层面上的打分:在接收到用户输入的一个查询指令后,需要经过表达模板库108的表达模板匹配,得到所述查询指令中存在于所述表达模板库108中的表达模板片段,例如用户输入“上海大众朗逸报价”, 则通过用户模板识别出查询指令中存在的“XX报价”模板。 It is worth mentioning: In a preferred embodiment of the present invention, in addition to the knowledge base to calculate the overall demand score, the intention in the analysis of demand, will also consider the level of expression on the template scoring: Upon receipt of a query entered by the user after the command, go through the expression template matching expression template database 108 to obtain the query instruction template fragments present in the template library 108 expression in the expression, for example, the user inputs "Sunny Shanghai Volkswagen Quote", the template identified by the user query command exists "XX offer" template. 在根据上述方法获得知识库整体需求得分的同时,查询指令又符合用户需求模板,表达模板库108也对所述查询指令在表达模板层面上进行打分,得到表达模板得分,则整体查询指令的需求强度得分为知识库整体需求得分与表达模板得分的加权之和,若该加权之和大于设定的阈值,则已加权之和最大的需求类型作为查询指令的需求意图,并根据所述需求意图在所述查询指令中加入相应的tag信息,例如“商品”、“新闻”、“图片”等。 While obtaining knowledge score overall demand by the above method, but also with user demand query instruction template, the template library 108 also expressing the query instructions on the template scoring expression level, resulting in the expression template scores, the overall query requirement command Knowledge intensity score of the score and the expression template overall demand weighted sum of the score, and if the weighted sum is greater than the set threshold, and the maximum weighted sum already needs demand type intended as query command, according to the needs and intentions adding the appropriate tag information in the query instructions, such as "commodity", "News", "picture" and so on.

[0185] 所述结构分类模块104用于结合所述知识库106,对UI模块102输入的查询指令进行智能化变换后发送至所述搜索模块105,其中,所述智能化变换即是语义内容扩充,所述的语义内容扩充包括了语义内容的归一化,以及语义类别的扩展。 After [0185] The structure of the classification module 104 is used in conjunction with knowledge base 106, the query instruction input module UI 102 is transmitted to the intelligent conversion search module 105, wherein the intelligent transformation that is the semantic content extension, said extension comprising semantic content normalized, and the expansion of the semantic content of the semantic class. 具体的,在所述查询指令有同属关系(上位属性的知识片段+下位属性的知识片段)时,例如,所述查询指令为“手机诺基亚”,此时,所述结构分类模块104在发送给所述搜索模块105时,即会在“手机” 这个上位属性的知识片段上加入“可以丢弃”的tag,这样,在搜索模块105对所述查询指令进行搜索时,即可通过“手机诺基亚”进行搜索,也可通过“诺基亚”进行搜索,且还可认为带有“手机诺基亚”文本信息的网页和只带有“诺基亚”文本信息的网页权值一样;另外,例如:若所述查询指令为“诺基亚”,则所述结构分类模块104在发送给所述搜索模块105时, 还可根据“诺基亚”对应其上位属性进行扩展,如扩展为“手机”,这样,在搜索模块105对“诺基亚”进行搜索时,可根据结果的数量判断是否需要扩展到“手机”进行搜索,如通过“诺基亚 Specifically, when the query instructions belong have the relationship (+ property knowledge fragment upper lower segment attribute knowledge), for example, the query command is "Nokia" In this case, the structure 104 to a classification module the search module 105, which will be added in the knowledge fragment "mobile phone" in this upper property "may discard" tag, you so that, when the search module 105 searches the query instruction, to the "Nokia" search, can also be searched by "Nokia" and may also be considered with the "Nokia" web page and only with the "Nokia" the same page weights text information text information; in addition, for example: if the query command the "Nokia", then the structure of the classification module 104 to transmit the search module 105, may be expanded according to "Nokia" attributes corresponding to its upper position, such as extended as "mobile phone", so that, in the search module 105 on the " Nokia "when you search, the results may determine whether the number needs to be extended to the" mobile phone "searches, such as through the" Nokia 搜索到的结果数量较小时,即可扩展到“手机”;又如:若所述查询指令为“手机”,则所述结构分类模块104可将“手机”对应其同位属性进行扩展,如扩展为“电脑”,这样的扩展可用于广告的推广,如在搜索页面的右侧即可根据“手机”这个查询指令进行广告推广, 又可根据“电脑”这个查询指令进行广告推广;再如,若所述查询指令为“诺基亚”,则所述结构分类模块104在发送给所述搜索模块105时,还可根据“诺基亚”对应其下位属性进行扩展,如扩展为“N71”、“N72”等,这样,在搜索模块105在搜索到带有“N71 ”、“N72”等文本信息网页时,也可根据这些网页的权值,判断是否作为搜索结果输出。 When a small number of search results can be extended to the "phone"; Another example: if the query instruction is a "mobile phone", then the configuration module 104 may be classified "mobile phone" which corresponds to the same bit extended attributes, such as extended the "computer", such extensions can be used to promote advertising, such as advertising can be carried out according to the "mobile phone" in the search query instruction to the right of the page, but also for advertising based on "computer" this query command; Again, If the query command is "Nokia", then the structure of the classification module 104 is transmitted to the search module 105, may be expanded according to "Nokia" which corresponds to the lower properties, such as extended as "N71", "N72" etc. Thus, the search module 105 to search, "N72" pages with text information such as "N71", also according to the weight of these pages, determines whether the output as a search result. 这种权值判断可参现有搜索引擎中的权值判断,在此不再赘述。 Such weight determination reference value may be determined in conventional search engines weights, are not repeated here. 总而言之,对所述查询指令进行扩展时,可根据搜索的策略,即可扩展其上位属性的知识片段,也可扩展其同位属性的知识片段,也可扩展其下位属性的知识片段。 In summary, when the extended query command, according to a search strategy, which can extend the upper property knowledge fragment, which can also be extended property knowledge parity fragments, fragments can also be extended their knowledge of the properties lower.

[0186] 所述搜索模块105用于接收经过所述需求意图分析模块103或所述结构分类模块104的智能化变换后的查询指令,并将该查询指令在网页存储库110中搜索,以得到搜索结果,同时,所述搜索模块105还用于将所述搜索结果返回至所述UI模块102。 [0186] The search module 105 for receiving the structural classification module 103 or the needs of the intended analysis module 104 queries the intelligent conversion instruction, and the instruction a search query page repository 110, to give search result, while the search module 105 is further configured to return the search result to the UI module 102.

[0187] 所述知识库106用于存储先验知识。 [0187] The knowledge base 106 for storing a priori knowledge. 在本发明最佳实施方式中,所述知识库主要存储为树形结构,对每一类知识库构建一棵知识库树,通过该知识库树的父亲节点标识其上位属性,右兄弟节点表示其同位属性,左兄弟节点表示其下位属性,如此迭代,直至叶子节点。 In the preferred embodiment of the invention, the main memory tree structure knowledge, to build a knowledge base for each category tree knowledge base, identify its parent node through the upper property knowledge tree, right sibling node represents its parity property, said its sibling left lower property, so iterations, until a leaf node. 如图6所示,“大众”为其最上位属性;其下位属性为“上海大众”;与所述“上海大众” 同位的有“一汽大众”;在所述“上海大众”下位的有“朗逸”,与所述“朗逸”同位的有“途 "Mass" to its uppermost property shown in FIG. 6; the lower attribute is "Shanghai Volkswagen"; a "FAW public" and the "Shanghai Volkswagen" with the bit; have the "Shanghai Volkswagen" the lower " Sunny ", and the" Sunny "bits have the same" way

观”......这种知识库的构建方法,本领域的普通技术人员可参现有技术完成,在此不再赘述。 View "...... method of constructing such a knowledge base, those of ordinary skill in the prior art reference may be complete, are not repeated here.

[0188] 用户历史行为库107用于存储用户历史搜索记录。 [0188] user behavior history database 107 is used to store user search history records. 优选地,其可包括查询指令、查询次数,以及加权点击数等信息。 Preferably, it may include information query instruction, queries, and the weighting hits like.

[0189] 表达模板挖掘模块108用于根据所述知识库106中的知识片段和所述用户历史行为库107中的用户历史查询指令,挖掘出表达模板,并将所述表达模板存储于所述表达模板库109中。 [0189] Expression template mining module 108 according to the pieces of knowledge in the knowledge base 106 and the user behavior history of the user history database 107 query instruction, excavated expression template, and the template stored in the expression expression template library 109. 相同需求的用户,在表达方式上会出现相似性,所述表达模板是指,一般用户在有一定查询需求时,其输入的查询指令为何,例如,当用户在查询汽车相关信息时,表达方式会有:“速腾怎么样”、“马六动力如何”等,其中即可抽取出“【汽车品牌/型号】怎么样”、“【汽车品牌/型号】动力如何”等表达汽车需求时常用的表达模板。 The same user demand, there will be a similarity in the expression, the expression refers to a template, generally when the user needs a certain query, query instruction input why, for example, when the user information query automobile, expression there will be: "Sagitar how", "how to power six horses," and so on, which can be extracted "[car make / model] how", "[car make / model] how to power" and other commonly used expression when demand for cars expression template. 在本发明的一实施方式中,具体为:首先在所述用户历史行为库107中包含知识库106知识片段的查询指令抽取出来,如在“马六如何”、“斯柯达如何”、“速腾如何”的查询指令中,抽取出知识片段:“马六”、“斯柯达”、“速腾”,其次将知识库片段替换成“【汽车品牌/型号】”符号,即生成“【汽车品牌/型号】如何”的候选表达模板;再次,统计生成的候选表达模板符合的知识库片段的数量,若该数量大于设定的阈值,则将所述候选表达模板作为表达模板,存于所述表达模板库109中;若该数量小于设定的阈值,则舍弃所述候选表达模板。 In one embodiment of the present invention, specifically as follows: First, a query command comprises 106 pieces of knowledge in the knowledge base in the user behavior history database 107 extracted, as described in "How to six horses", "how Skoda", "How suteng "query command, extract the pieces of knowledge:" six horses "," Skoda "," Sagitar ", followed by the replacement knowledge fragment as" [car make / model] "symbol, which generates" [car make / model] how "expression candidate template; again, the number of segments of the knowledge base to generate statistical template matching expression candidate, if the number is greater than a set threshold, the template will be expressed as an expression of the candidate template stored in the template library expression 109; if the number is less than a set threshold, the candidate is discarded expression template.

[0190] 所述表达模板库109用于存储由所述表达模板挖掘模块108挖掘出的表达模板。 [0190] The expression template in the gallery 109 for storing the template expression excavated mining module 108.

[0191] 所述网页存储库110用于存储网页数据和该网页数据的索引信息。 The [0191] page store 110 for storing index information of the web page data and page data. 该数据库即是普通搜索引擎常用的数据库,在此不再赘述。 The database that is commonly used in ordinary search engine database, which will not be repeated here.

[0192] 如图5所示,在本发明第四实施方式中,所述搜索引擎包括了web服务模块101、 与web服务模块101交互通信的UI模块102、与所述UI模块102通信的需求意图分析模块103、与所述UI模块102通信的结构分析模块104、与所述结构分析模块104通信的搜索模块105,以及与所述需求意图分析模块103、所述结构分析模块104交互通信的知识库106、与所述知识库106通信的用户历史行为库107、与所述知识库106、用户历史行为库107通信的表达模板挖掘模块108、与所述表达模板挖掘模块108和所述需求意图分析模块103通信的表达模板库109,以及与所述搜索模块105通信的网页存储库110、第一垂直搜索数据 [0192], in the fourth embodiment of the present invention, the search engine 5 includes web service module 101, the UI module 102 and the communication needs of web interaction with the UI module 101 in communication service module 102 intended analysis module 103, the UI module 102 and the communication structure analysis module 104, the search module 104 in communication structure analysis module 105, and the requirements and intended analysis module 103, the interactive communication structure analysis module 104 knowledge base 106, and the user behavior history database 106 in communication with the repository 107, and 106 the knowledge, the expression template mining module 107 communicating with the user behavior history database 108, and the expression of the template and excavation module 108 needs expression analysis template library module 103 is intended for communication 109 and a communication module 105 and the search web page repository 110, a first vertical search data

111a、第二垂直搜索数据库Illb........第N垂直搜索数据库llln。 111a, a second vertical search database Illb ........ N-th vertical search database llln. 值得一提的是,这些 It is worth mentioning that these

模块即可存储并运行于同一服务器中,也可存储并运行在多台服务器中。 Modules can be stored and run on the same server can also be stored and run on multiple servers.

[0193] 所述web服务模块101用于通过网络协议接收从客户端20传来的查询指令,并将该查询指令转到UI模块102,另外,该web服务模块101还用于接收所述UI模块102返回的结果页面,并将所述结果页面返回至客户端20。 [0193] The web service module 101 receives the query command 20 transmitted from a client through a network protocol, and the query instruction to the UI module 102, Further, the web service module 101 is further configured to receive the UI module 102 returns the result page, and the results page returned to the client 20.

[0194] 所述UI模块102用于接收所述web服务模块101传送的查询指令,并将该查询指令发送至所述查询指令分析模块103 ;另外,所述UI模块102还用于接收所述搜索模块104 返回的搜索结果,并将所述搜索结果拼装为结果页面后,返回所述结果页面至所述web服务模块101。 [0194] The UI module 102 for receiving the web service module 101 transmits a query instruction, and sends the inquiry command to the query command analysis module 103; Further, the UI module 102 is further configured to receive the search module 104 returns the search results, and the search results page after assembly as a result, returns the results page to the web service module 101.

[0195] 所述需求意图分析模块103用于调用所述知识库106、用户历史行为库107,以及所述表达模板库108,以对接收到的查询指令进行需求意图分析,明确所述查询指令的需求意图。 [0195] The intent demand analysis module 103 is used to call the Knowledge Base 106, user behavior history database 107, and the expression template library 108, to check the received instruction needs Intent Analysis, clear the query command It demands intentions. 在本发明中,所述意图分析模块103首先通过所述用户历史行为库107给所述知识库106中的各个知识片段的各个需求意图打分,具体的:用户在查询某类需求时,会点击相应的满足他需求的结果,如用户想获得汽车报价相关的信息,在搜索引擎输入查询指令“朗逸”后,会点击搜索引擎返回的汽车网站的tol,如“网易车库”,此时用户输入的查询指令“朗逸”片段以及用户点击的tol “网易车库”都隐含反映用户有找车相关信息的需求,基于这一点,本发明在计算各个知识片段的需求意图时,根据某个知识片段点击某类tol的数目/某个知识片段点击的总tol,来确定此知识片段需求意图的得分,如在所述用户历史行为库107中得知,查询指令为“朗逸”,其点击的总Url数为10个,其中,商品类tol为5 个,新闻类Url为3个,图片类Url为2个,则可计算出该查询指令 In the present invention, the analysis module 103 first intention by the user behavior history database to the individual requirements in each of the pieces of knowledge in the knowledge base 106 intention score 107, specifically: a user query certain requirements, will click corresponding to satisfy his demands results, such as the user wants to get car quotes related information, enter the query command in the search engine "Sunny" will click tol search engine returns automotive Web site, such as "Netease garage", then the user input query command "Sunny" fragment and tol user clicks the "NetEase garage" implicitly reflect user needs to find car-related information, based on this, the present invention in the calculation of the various pieces of knowledge of the intention of the demand, according to some pieces of knowledge tol certain number of clicks / click on a fragment of the total tol knowledge, this knowledge to determine the needs of the intention of scoring fragments, known as the user behavior history database 107, the query command is "Sunny", the click of the total Url number 10, wherein the commodity is tol 5, three news Url, Url picture type is 2, the query command can be calculated 商品类的需求意图为0. 5,新闻类的需求意图为0. 3,图片类的需求意图为0. 2 ;其次,在接收到用户输入的一个查询指令后,需要经过知识库106的知识匹配,得到所述查询指令中存在于所述知识库106 中的知识片段,并综合计算所述查询指令的知识库整体需求强度。 Commodity demand intention is 0.5, the demand for news is intent 0.3, demand is intent picture category 0.2; secondly, after receiving a query command entered by the user, need to go through knowledge repository 106 matching the query to obtain the strength of the overall demand instruction is present in the knowledge base knowledge base knowledge segment 106, and calculates the integrated query command. 例如用户输入“上海大众朗逸报价”,则通过知识库可获得“上海大众” “朗逸”的知识片段。 For example, the user inputs "Shanghai Volkswagen Lavida quote," through the knowledge base "Shanghai Volkswagen", "Sunny" fragments of knowledge available. 获得初步信息后,首先将知识片段“上海大众”和“朗逸”各自的需求意图得分加总,得到第一分数,其次,再通过知识片段“上海大众”与“朗逸”的关系,加减所述第一分数,得到知识库整体需求得分, 在本发明最佳实施方式中,若知识片段的关系为同属关系,则加分;若知识片段为非同属关系,则减分;若该知识库整体需求得分大于设定的阈值,则以知识库整体需求得分最高的需求类型作为查询指令的需求意图,并根据所述需求意图在所述查询指令中加入相应的tag 信息,例如“商品”、“新闻”、“图片”等。 After obtaining preliminary information, first of all the pieces of knowledge "Shanghai Volkswagen" and "Sunny" their needs intention score summed to obtain a first fraction, and secondly, through the pieces of knowledge Relations "Shanghai Volkswagen" and "Sunny", the subtraction of the said first fraction, to obtain the overall score needs knowledge, in the preferred embodiment of the present invention, if the relationship between knowledge fragments belong to the relationship, the points; if knowledge fragments belong to the non-relationship, it is less points; if the repository overall demand score is greater than a set threshold, places the highest scoring overall demand knowledge requirement types intended as query command, and the tag information added to the corresponding instruction in the query according to the intended requirements, e.g., "merchandise", "News", "picture" and so on. 值得一提的是:在本发明的最佳实施方式中,除了计算知识库整体需求得分外,在分析需求意图时,还会考虑表达模板层面上的打分:在接收到用户输入的一个查询指令后,需要经过表达模板库108的表达模板匹配,得到所述查询指令中存在于所述表达模板库108中的表达模板片段,例如用户输入“上海大众朗逸报价”, 则通过用户模板识别出查询指令中存在的“XX报价”模板。 It is worth mentioning: In a preferred embodiment of the present invention, in addition to the knowledge base to calculate the overall demand score, the intention in the analysis of demand, will also consider the level of expression on the template scoring: Upon receipt of a query entered by the user after the command, go through the expression template matching expression template database 108 to obtain the query instruction template fragments present in the template library 108 expression in the expression, for example, the user inputs "Sunny Shanghai Volkswagen Quote", the template identified by the user query command exists "XX offer" template. 在根据上述方法获得知识库整体需求得分的同时,查询指令又符合用户需求模板,表达模板库108也对所述查询指令在表达模板层面上进行打分,得到表达模板得分,则整体查询指令的需求强度得分为知识库整体需求得分与表达模板得分的加权之和,若该加权之和大于设定的阈值,则已加权之和最大的需求类型作为查询指令的需求意图。 While obtaining knowledge score overall demand by the above method, but also with user demand query instruction template, the template library 108 also expressing the query instructions on the template scoring expression level, resulting in the expression template scores, the overall query requirement command strength score score score expression template weighted sum of overall demand knowledge, if the weighted sum is greater than the set threshold, and the maximum weighted sum already needs demand type intended as query command.

[0196] 所述结构分类模块104用于结合所述知识库106,对UI模块102输入查询指令进行智能化变换后发送至所述搜索模块105,其中,所述智能化变换即是语义内容扩充,所述的语义内容扩充包括了语义内容的归一化,以及语义类别的扩展。 [0196] The classification module 104 is used in conjunction with the structure of the knowledge base 106, the search module 105 transmits to the input of the UI module 102 intelligently query instruction transformation, wherein the transformation that is intelligent expansion semantic content , the semantic content of the normalization comprises expanded, and the expansion of the semantic content of the semantic class. 具体的,在所述查询指令有同属关系(上位属性的知识片段+下位属性的知识片段)时,例如,所述查询指令为“手机诺基亚”,此时,所述结构分类模块104在发送给所述搜索模块105时,即会在“手机”这个上位属性的知识片段上加入“可以丢弃”的tag,这样,在搜索模块105对所述查询指令进行搜索时,即可通过“手机诺基亚”进行搜索,也可通过“诺基亚”进行搜索,且还可认为带有“手机诺基亚”文本信息的网页和只带有“诺基亚”文本信息的网页权值一样;另外,例如:若所述查询指令为“诺基亚”,则所述结构分类模块104在发送给所述搜索模块105时,还可根据“诺基亚”对应其上位属性进行扩展,如扩展为“手机”,这样,在搜索模块105对“诺基亚” 进行搜索时,可根据结果的数量判断是否需要扩展到“手机”进行搜索,如通过“诺基亚 Specifically, when the query instructions belong have the relationship (+ property knowledge fragment upper lower segment attribute knowledge), for example, the query command is "Nokia" In this case, the structure 104 to a classification module the search module 105, which will be added in the knowledge fragment "mobile phone" in this upper property "may discard" tag, you so that, when the search module 105 searches the query instruction, to the "Nokia" search, can also be searched by "Nokia" and may also be considered with the "Nokia" web page and only with the "Nokia" the same page weights text information text information; in addition, for example: if the query command the "Nokia", then the structure of the classification module 104 to transmit the search module 105, may be expanded according to "Nokia" attributes corresponding to its upper position, such as extended as "mobile phone", so that, in the search module 105 on the " Nokia "when you search, the results may determine whether the number needs to be extended to the" mobile phone "searches, such as through the" Nokia 搜索到的结果数量较小时,即可扩展到“手机”;又如:若所述查询指令为“手机”,则所述结构分类模块104可将“手机”对应其同位属性进行扩展,如扩展为“电脑”,这样的扩展可用于广告的推广,如在搜索页面的右侧即可根据“手机”这个查询指令进行广告推广,又可根据“电脑”这个查询指令进行广告推广;再如,若所述查询指令为“诺基亚”,则所述结构分类模块104在发送给所述搜索模块105时,还可根据“诺基亚”对应其下位属性进行扩展,如扩展为“N71”、“N72”等,这样,在搜索模块105在搜索到带有“N71”、“N72”等文本信息网页时,也可根据这些网页的权值,判断是否作为搜索结果输出。 When a small number of search results can be extended to the "phone"; Another example: if the query instruction is a "mobile phone", then the configuration module 104 may be classified "mobile phone" which corresponds to the same bit extended attributes, such as extended the "computer", such extensions can be used to promote advertising, such as advertising can be carried out according to the "mobile phone" in the search query instruction to the right of the page, but also for advertising based on "computer" this query command; Again, If the query command is "Nokia", then the structure of the classification module 104 is transmitted to the search module 105, may be expanded according to "Nokia" which corresponds to the lower properties, such as extended as "N71", "N72" etc. Thus, the search module 105 to search, "N72" pages with text information such as "N71", also according to the weight of these pages, determines whether the output as a search result. 这种权值判断可参现有搜索引擎中的权值判断,在此不再赘述。 Such weight determination reference value may be determined in conventional search engines weights, are not repeated here. 总而言之,对所述查询指令进行扩展时,可根据搜索的策略,即可扩展其上位属性的知识片段,也可扩展其同位属性的知识片段,也可扩展其下位属性的知识片段。 In summary, when the extended query command, according to a search strategy, which can extend the upper property knowledge fragment, which can also be extended property knowledge parity fragments, fragments can also be extended their knowledge of the properties lower.

[0197] 所述搜索模块105用于接收经过所述需求意图分析模块103或所述结构分类模块104的智能化变换后的查询指令,并将该查询指令在多个垂直搜索数据库(第一垂直搜索 [0197] The search module 105 for receiving the demands of the intended structure analysis module 103 or the classification module 104 queries the intelligent conversion instruction, and the instruction in the plurality of vertical search query databases (first vertical search for

数据库111a、第二垂直搜索数据库Illb........第N垂直数据库llln)的其中之一,以及 Database 111a, a second vertical search database Illb ........ N-th vertical database llln) wherein one, and

所述网页存储库1110中搜索,以得到搜索结果,同时,所述搜索模块105还用于将所述搜索结果返回至所述UI模块102。 The web page repository 1110 searches, to obtain search results, and the search module 105 is further configured to return the search result to the UI module 102. 值得一提的是:选择某个垂直搜索数据库进行垂直搜索是通过查询指令的需求意图确定的,例如:若查询指令的需求意图为“商品”,则在商品垂直搜索数据库中进行搜索;所查询指令的需求意图为“图片”,则在图片垂直搜索数据库中进行搜索,其中,在垂直搜索数据库中搜索到的一条或多条结果,会插入至在网页存储库中搜索到的结果中,形成整体搜索结果。 It is worth mentioning: select a vertical search vertical search databases by querying the command is intent demand determined, for example: If the query command is intended to demand "commodity", the search vertical search in the commodity database; the query demand intended instructions is "image", the search is performed in the vertical image search database, wherein, in the searched vertical search results database and one or more, is inserted into the page to the search result repository, the formation of the overall search results. 所述垂直搜索,即是在某个特定的类别下进行搜索,其具体的搜索方法和系统在本领域中已多有现有技术揭示,在此不再赘述。 The vertical search that is carried out in the search for a specific category, the specific search method and system of the present art has revealed more than a prior art, it is not repeated here.

[0198] 所述知识库106用于存储先验知识。 [0198] The knowledge base 106 for storing a priori knowledge. 在本发明最佳实施方式中,所述知识库主要存储为树形结构,对每一类知识库构建一棵知识库树,通过该知识库树的父亲节点标识其上位属性,右兄弟节点表示其同位属性,左兄弟节点表示其下位属性,如此迭代,直至叶子节点。 In the preferred embodiment of the invention, the main memory tree structure knowledge, to build a knowledge base for each category tree knowledge base, identify its parent node through the upper property knowledge tree, right sibling node represents its parity property, said its sibling left lower property, so iterations, until a leaf node. 如图6所示,“大众”为其最上位属性;其下位属性为“上海大众”;与所述“上海大众” 同位的有“一汽大众”;在所述“上海大众”下位的有“朗逸”,与所述“朗逸”同位的有“途观”......这种知识库的构建方法,本领域的普通技术人员可参现有技术完成,在此不再赘述。 "Mass" to its uppermost property shown in FIG. 6; the lower attribute is "Shanghai Volkswagen"; a "FAW public" and the "Shanghai Volkswagen" with the bit; have the "Shanghai Volkswagen" the lower " Sunny ", and the" "have the same position" Sunny Tiguan "...... method of constructing such a knowledge base, those of ordinary skill in the prior art reference may be complete, are not repeated here.

[0199] 用户历史行为库107用于存储用户历史搜索记录。 [0199] user behavior history database 107 is used to store user search history records. 优选地,其可包括查询指令、查询次数,以及加权点击数等信息。 Preferably, it may include information query instruction, queries, and the weighting hits like.

[0200] 表达模板挖掘模块108用于根据所述知识库106中的知识片段和所述用户历史行为库107中的用户历史查询指令,挖掘出表达模板,并将所述表达模板存储于所述表达模板库109中。 [0200] Expression template mining module 108 according to the pieces of knowledge in the knowledge base 106 and the user behavior history of the user history database 107 query instruction, excavated expression template, and the template stored in the expression expression template library 109. 相同需求的用户,在表达方式上会出现相似性,所述表达模板是指,一般用户在有一定查询需求时,其输入的查询指令为何,例如,当用户在查询汽车相关信息时,表达方式会有:“速腾怎么样”、“马六动力如何”等,其中即可抽取出“【汽车品牌/型号】怎么样”、“【汽车品牌/型号】动力如何”等表达汽车需求时常用的表达模板。 The same user demand, there will be a similarity in the expression, the expression refers to a template, generally when the user needs a certain query, query instruction input why, for example, when the user information query automobile, expression there will be: "Sagitar how", "how to power six horses," and so on, which can be extracted "[car make / model] how", "[car make / model] how to power" and other commonly used expression when demand for cars expression template. 在本发明的一实施方式中,具体为:首先在所述用户历史行为库107中包含知识库106知识片段的查询指令抽取出来,如在“马六如何”、“斯柯达如何”、“速腾如何”的查询指令中,抽取出知识片段:“马六”、“斯柯达”、“速腾”,其次将知识库片段替换成“【汽车品牌/型号】”符号,即生成“【汽车品牌/型号】如何”的候选表达模板;再次,统计生成的候选表达模板符合的知识库片段的数量,若该数量大于设定的阈值,则将所述候选表达模板作为表达模板,存于所述表达模板库109中;若该数量小于设定的阈值,则舍弃所述候选表达模板。 In one embodiment of the present invention, specifically as follows: First, a query command comprises 106 pieces of knowledge in the knowledge base in the user behavior history database 107 extracted, as described in "How to six horses", "how Skoda", "How suteng "query command, extract the pieces of knowledge:" six horses "," Skoda "," Sagitar ", followed by the replacement knowledge fragment as" [car make / model] "symbol, which generates" [car make / model] how "expression candidate template; again, the number of segments of the knowledge base to generate statistical template matching expression candidate, if the number is greater than a set threshold, the template will be expressed as an expression of the candidate template stored in the template library expression 109; if the number is less than a set threshold, the candidate is discarded expression template.

[0201] 所述表达模板库109用于存储由所述表达模板挖掘模块108挖掘出的表达模板。 [0201] The expression template in the gallery 109 for storing the template expression excavated mining module 108.

[0202] 所述网页存储库110用于存储网页数据和该网页数据的索引信息。 The [0202] page store 110 for storing index information of the web page data and page data. 该数据库即是普通搜索引擎常用的数据库,在此不再赘述。 The database that is commonly used in ordinary search engine database, which will not be repeated here.

[0203] 所述第一垂直搜索数据库111a、第二垂直搜索数据库Illb........第N垂直搜索 [0203] The first vertical search database 111a, a second vertical search database Illb ........ vertical search of N

数据库Illn用于存储特定类别数据和该特定类别数据的索引信息,例如商品数据、商品索引;新闻数据、新闻索引;图片数据、图片索引等。 Illn index database for storing information about specific categories of data and the specific categories of data, such as product data, commodity index; news, news index; picture data, images and other indexes.

[0204] 如图7所示,本发明第一实施方式的搜索方法包括以下步骤: [0204] As shown in FIG 7, the first embodiment of the search method of the present invention comprises the steps of:

[0205] Si、接收查询指令;优选地,该查询指令是用户通过客户端上的浏览器输入的至web服务模块101,该web服务模块101在得到所述查询命令后,会将该查询命令转到UI模块102 ; [0205] Si, receiving a query instruction; preferably, the query instruction is the user service module 101 to the web browser on the client input, the web service module 101 after receiving the query, the query command will go to the UI module 102;

[0206] S2、基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图;优选地,该步骤是通过所述需求意图分析模块103完成的; [0206] S2, based on knowledge of the intended query instruction needs analysis, clear the query requirement command is intended; preferably, this step is intended by the demand analysis module 103 is completed;

[0207] S3、将带有需求意图的所述查询指令在数据库中搜索,得到搜索结果;优选地,该步骤是通过所述搜索模块105完成的; The query instruction [0207] S3, with the intention of searching the database needs to obtain search results; Preferably, this step is performed by the search module 105 of completion;

[0208] S4、输出所述搜索结果。 [0208] S4, outputs the search result. 优选地,该步骤是在所述UI模块102和所述web服务模块101中完成的,搜索结果从所述搜索模块104返回至所述UI模块102,并通过所述UI模块102将所述搜索结果拼装为结果页面后,返回所述结果页面至所述web服务模块101,从而通过所述web服务模块101返回至客户端浏览器。 Preferably, this step is done in the UI module 102 and the web service module 101, the search results returned from the search module 104 to the UI module 102, and the UI module 102 through the search results the results for the assembled page, returning the result page to the web service module 101, thereby returning to the client browser via the web service module 101.

[0209] 其中,在所述S3步骤中的数据库即可为网页存储库110,或与需求意图相对应的垂直搜索数据库;当然,也可包括网页存储库110和与所述需求意图相对应的垂直搜索数据库。 [0209] wherein, in the step S3 in the database to the Web page repository 110, corresponding to the intention or demand vertical search database; of course, may also include a web page store 110 correspond to the needs and intentions vertical search database.

[0210] 在所述S2步骤和S3步骤之间,还包括语义扩充步骤: [0210] between the step S2 and the step S3, further comprising the step of semantic extension:

[0211] 基于所述知识库对接收到的查询指令进行语义扩充;优选地,该步骤是通过结构分析模块104完成的。 [0211] Knowledge of the query instruction based on the received extension semantics; Preferably, this step is performed by the analysis module 104 completes the structure. [0212] 如图8所示,本发明第二实施方式的搜索方法包括以下步骤: [0212] As shown, the searching method of the second embodiment of the present invention includes the steps 8:

[0213] Si'、接收查询指令;优选地,该查询指令是用户通过客户端上的浏览器输入的至web服务模块101,该web服务模块101在得到所述查询命令后,会将该查询命令转到UI模块102 ; [0213] Si ', receiving a query instruction; preferably, the query instruction is the user service module 101 to the web browser on the client's input, the web service module 101 after receiving the query, the query will command to the UI module 102;

[0214] S2'、基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图,同时,基于所述知识库对接收到的查询指令进行语义扩充;优选地,该步骤是通过所述需求意图分析模块103和所述结构分析模块104完成的; [0214] S2 ', based on knowledge of the query instructions intended analysis requirements, the query requirement clear intention command, while semantic query expansion based on the instruction received knowledge base; preferably, this step It is intended by the demand analysis module 103 and the analysis module 104 completes the structure;

[0215] S3'、将带有需求意图并扩充语义的查询指令在数据库中搜索,得到搜索结果;优选地,该步骤是通过所述搜索模块105完成的; [0215] S3 ', the intent and requirements with the semantic query expansion instruction searches the database to obtain search results; Preferably, this step is accomplished by the search module 105;

[0216] S4'、输出所述搜索结果。 [0216] S4 ', the output of the search result. 优选地,该步骤是在所述UI模块102和所述web服务模块101中完成的,搜索结果从所述搜索模块104返回至所述UI模块102,并通过所述UI模块102将所述搜索结果拼装为结果页面后,返回所述结果页面至所述web服务模块101,从而通过所述web服务模块101返回至客户端浏览器。 Preferably, this step is done in the UI module 102 and the web service module 101, the search results returned from the search module 104 to the UI module 102, and the UI module 102 through the search results the results for the assembled page, returning the result page to the web service module 101, thereby returning to the client browser via the web service module 101.

[0217] 其中,在所述S3'步骤中的数据库即可为网页存储库110,或与需求意图相对应的垂直搜索数据库;当然,也可包括网页存储库110和与所述需求意图相对应的垂直搜索数据库。 [0217] wherein, in the database S3 'step 110 to store a Web page, or a corresponding demand intended vertical search database; of course, may also include a web page store 110 and are intended to correspond to the needs vertical search database.

[0218] 如图9所示,本发明第三实施方式的搜索方法包括以下步骤: [0218] As shown, the searching method of the third embodiment of the present invention comprises the following 9 steps:

[0219] S10、接收查询指令;优选地,该查询指令是用户通过客户端上的浏览器输入的至web服务模块101,该web服务模块101在得到所述查询命令后,会将该查询命令转到UI模块102 ; [0219] S10, receiving a query instruction; preferably, the query instruction is the user service module 101 to the web browser on the client's input, the web service module 101 after receiving the query, the query command will go to the UI module 102;

[0220] S20、基于知识库和表达模板库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图;优选地,该步骤是通过所述需求意图分析模块103完成的; [0220] S20, the knowledge base, and the query expression template library instruction needs analysis based on intention, the query requirement expressly intended instruction; preferably, this step is intended by the demand analysis module 103 is completed;

[0221] S30、将带有需求意图的所述查询指令在数据库中搜索,得到搜索结果;优选地,该步骤是通过所述搜索模块105完成的; The query instruction [0221] S30, the intent with the demand search the database to obtain search results; Preferably, this step is performed by the search module 105 of completion;

[0222] S40、输出所述搜索结果。 [0222] S40, the output of the search result. 优选地,该步骤是在所述UI模块102和所述web服务模块101中完成的,搜索结果从所述搜索模块104返回至所述UI模块102,并通过所述UI模块102将所述搜索结果拼装为结果页面后,返回所述结果页面至所述web服务模块101,从而通过所述web服务模块101返回至客户端浏览器。 Preferably, this step is done in the UI module 102 and the web service module 101, the search results returned from the search module 104 to the UI module 102, and the UI module 102 through the search results the results for the assembled page, returning the result page to the web service module 101, thereby returning to the client browser via the web service module 101.

[0223] 其中,在所述S30步骤中的数据库即可为网页存储库110,或与需求意图相对应的垂直搜索数据库;当然,也可包括网页存储库110和与所述需求意图相对应的垂直搜索数据库。 [0223] wherein, in the step S30 in the database to the Web page repository 110, corresponding to the intention or demand vertical search database; of course, may also include a web page store 110 correspond to the needs and intentions vertical search database.

[0224] 在所述S20步骤和S30步骤之间,还包括语义扩充步骤: [0224] In the step S20 between step S30 and, further comprising the step of semantic extension:

[0225] 基于所述知识库对接收到的查询指令进行语义扩充;优选地,该步骤是通过结构分析模块104完成的。 [0225] Knowledge of the query instruction based on the received extension semantics; Preferably, this step is performed by the analysis module 104 completes the structure.

[0226] 如图8所示,本发明第四实施方式的搜索方法包括以下步骤: [0226] As shown, the searching method of the fourth embodiment of the present invention includes the steps 8:

[0227] S10'、接收查询指令;优选地,该查询指令是用户通过客户端上的浏览器输入的至web服务模块101,该web服务模块101在得到所述查询命令后,会将该查询命令转到UI模块102 ; [0227] S10 ', receiving a query instruction; preferably, the query instruction is the user service module 101 to the web browser on the client's input, the web service module 101 after receiving the query, the query will command to the UI module 102;

[0228] S20'、基于知识库和表达模板库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图,同时,基于所述知识库对接收到的查询指令进行语义扩充;优选地,该步骤是通过所述需求意图分析模块103和所述结构分析模块104完成的; [0228] S20 ', the knowledge base, and the query expression template library instruction needs analysis is intended, the requirements are intended to define the query instruction, while semantic query expansion based on the instruction based on the received knowledge base; preferably , this step is intended by the demand analysis module 103 and the analysis module 104 completes the structure;

[0229] S30'、将带有需求意图并扩充语义的查询指令在数据库中搜索,得到搜索结果;优选地,该步骤是通过所述搜索模块105完成的; [0229] S30 ', the intent and requirements with expanded search query instruction semantics database to obtain search results; Preferably, this step is accomplished by the search module 105;

[0230] S40,、输出所述搜索结果。 [0230] S40 ,, the output of the search result. 优选地,该步骤是在所述UI模块102和所述web服务模块101中完成的,搜索结果从所述搜索模块104返回至所述UI模块102,并通过所述UI 模块102将所述搜索结果拼装为结果页面后,返回所述结果页面至所述web服务模块101, 从而通过所述web服务模块101返回至客户端浏览器。 Preferably, this step is done in the UI module 102 and the web service module 101, the search results returned from the search module 104 to the UI module 102, and the UI module 102 through the search results the results for the assembled page, returning the result page to the web service module 101, thereby returning to the client browser via the web service module 101.

[0231] 其中,在所述S30'步骤中的数据库即可为网页存储库110,或与需求意图相对应的垂直搜索数据库;当然,也可包括网页存储库110和与所述需求意图相对应的垂直搜索数据库。 [0231] wherein, in said database S30 'to step 110 to store the page, or a corresponding demand intended vertical search database; of course, may also include a web page store 110 and are intended to correspond to the needs vertical search database.

[0232] 如图11所示,在本发明第一实施方式、第二实施方式、第三实施方式、第四实施方式的搜索方法中,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”步骤的一实施方式,包括以下流程: [0232] 11, in a first embodiment of the present invention, the second embodiment, the third embodiment embodiment, the search method of the fourth embodiment, the "Knowledge-based instruction needs the query intent analysis, clear intent of the query instruction needs "a way of implementation steps, including the following process:

[0233] S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;具体的:用户在查询某类需求时,会点击相应的满足他需求的结果,如用户想获得汽车报价相关的信息,在搜索引擎输入查询指令“朗逸”后,会点击搜索引擎返回的汽车网站的tol,如“网易车库”,此时用户输入的查询指令“朗逸”片段以及用户点击的Url “网易车库”都隐含反映用户有找车相关信息的需求,基于这一点,本发明在计算各个知识片段的需求意图时,根据某个知识片段点击某类tol的数目/某个知识片段点击的总此1,来确定此知识片段需求意图的得分,如在所述用户历史行为库107中得知,查询指令为“朗逸”,其点击的总Url数为10个,其中,商品类to 1为5个, 新闻类Url为3个,图片类Url为2个,则可计算出该查询 [0233] S200, user behavior history database to individual needs of each segment of the knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring; specifically: the demand for certain types of queries when users will click on the appropriate to meet his demands results, such as the user wants to get car quotes related information, enter the query command in the search engine "Sunny" will click tol search engine returns automotive Web site, such as "Netease garage", then the user input query command "Sunny" fragment and Url "NetEase garage" are implicitly reflect the user clicks the user has to find the car related information needs, based on this, the present invention in the calculation of the various pieces of knowledge of the intention of the demand, according to some pieces of knowledge click this total 1, to determine the needs of this segment of knowledge of the intention of the number of certain types of tol / scores click of a fragment of knowledge, such as that in the user behavior history database 107, query instruction is "Sunny", the click of the total Url number 10, in which commodity to 1 to 5, for the three news Url, Url class picture for the two, you can calculate the query 令的商品类的需求意图为0. 5, 新闻类的需求意图为0. 3,图片类的需求意图为0. 2 ; Commodity demand intention of orders for the 0.5, demand intention news is 0.3, demand is intent picture category 0.2;

[0234] S201、在接收到用户输入的一个查询指令后,将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段;例如用户输入“上海大众朗逸报价”,则通过知识库可获得“上海大众” “朗逸”的知识片段; [0234] S201, after receiving an instruction input by a user query, the query command and the matching pieces of knowledge, to give at least a fragment of the query instructions knowledge matched; for example, the user enters "Sunny Shanghai Volkswagen Quote", "Shanghai Volkswagen", "Sunny" knowledge fragments can be obtained through the knowledge base;

[0235] S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数; 例如:知识片段“上海大众”和“朗逸”各自的需求意图得分加总,得到第一分数; [0235] S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first score; for example: pieces of knowledge "Shanghai public" and "Sunny" are each intended to demand the total score is added, to give first a score;

[0236] S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;在本发明最佳实施方式中,若知识片段的关系为同属关系,则加分;若知识片段为非同属关系,则减分; [0236] S203, the query command through matched pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score of the knowledge base; In a preferred embodiment of the present invention, If the relationship between knowledge fragments belong to the relationship, the points; if knowledge fragments belong to the non-relationship, it is less points;

[0237] S204、判断所述知识库整体需求得分是否大于一设定阈值; [0237] S204, determines whether or not the overall demand knowledge score greater than a predetermined threshold value;

[0238] S205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图; [0238] S205, if greater than the set threshold value, places the highest scoring overall demand knowledge of the demand type as a query instruction needs intention;

[0239] S206、若小于所述设定阈值,则判断所述查询指令无明显需求意图,按照普通搜索引擎搜索方式进行搜索,在此不再赘述。 [0239] S206, if less than the set threshold value, it is determined that the query instruction needs no intention, searching by search engine ordinary manner, not described herein again.

[0240] 如图12所示,在本发明第一实施方式、第二实施方式、第三实施方式、第四实施方式的搜索方法中,所述“基于知识库和表达模板库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”步骤的另一实施方式,包括以下流程: [0240] 12, in a first embodiment of the present invention, the second embodiment, the third embodiment embodiment, the search method of the fourth embodiment, the "Knowledge-based template library and the query expression instruction needs intent analysis, another embodiment of the clear intent of the query command needs "steps, including the following process:

[0241] S200'、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;具体的:用户在查询某类需求时,会点击相应的满足他需求的结果,如用户想获得汽车报价相关的信息,在搜索引擎输入查询指令“朗逸”后,会点击搜索引擎返回的汽车网站的tol,如“网易车库”,此时用户输入的查询指令“朗逸”片段以及用户点击的Url “网易车库”都隐含反映用户有找车相关信息的需求,基于这一点,本发明在计算各个知识片段的需求意图时,根据某个知识片段点击某类tol的数目/某个知识片段点击的总此1,来确定此知识片段需求意图的得分,如在所述用户历史行为库107中得知,查询指令为“朗逸”,其点击的总to 1数为10个,其中,商品类to 1为5个, 新闻类tol为3个,图片类tol为2个,则可计算出该查 [0241] S200 ', by the user's historical behavior to the individual needs of each library fragment knowledge of the intention of scoring Knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring; specifically: the demand for certain types of queries when users will click corresponding to satisfy his demands results, such as the user wants to get car quotes related information, enter the query command in the search engine "Sunny" will click tol search engine returns automotive Web site, such as "Netease garage", then the user input query command "Sunny" fragment and Url user clicks the "NetEase garage" implicitly reflect user needs to find car-related information, based on this, the present invention in the calculation of the various pieces of knowledge of the intention of the demand, according to some pieces of knowledge tol certain number of clicks / total 1 a knowledge of this segment clicks to determine this segment demand the knowledge of the intention of scoring, as known in the user behavior history database 107, query instruction is "Sunny", with its click total number of 10 to 1, wherein the commodity is 5 to 1, three news tol, tol picture type is 2, the search can be calculated 指令的商品类的需求意图为0. 5, 新闻类的需求意图为0. 3,图片类的需求意图为0. 2 ; Commodity demand intent instruction is 0.5, the demand for news is intent 0.3, demand is intent picture category 0.2;

[0242] S201'、在接收到用户输入的一个查询指令后,将所述查询指令与知识片段和存储于表达模板库中的表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一个表达模板;例如用户输入“上海大众朗逸报价”,则通过知识库可获得“上海大众” “朗逸”的知识片段;通过表达模板库克获得查询指令中存在的“XX报价”的表达模板; [0242] S201 ', after receiving a command input by a user query, the query expression template matching and instruction segments stored in the knowledge expression template library to obtain the fragment of the at least one knowledge query instruction matches and the expression of a template; for example, the user enters "Shanghai Volkswagen Lavida quote," through the knowledge base "Shanghai Volkswagen", "Sunny" of pieces of knowledge; obtain query instructions by the presence of the Cook expression template "available XX quote," the expression template ;

[0243] S202'、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;例如:知识片段“上海大众”和“朗逸”各自的需求意图得分加总,得到第一分数; [0243] S202 ', the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first score; for example: pieces of knowledge "Shanghai public" and "Sunny" are each intended to demand the total score is added to give The first score;

[0244] S203'、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;在本发明最佳实施方式中,若知识片段的关系为同属关系,则加分;若知识片段为非同属关系,则减分; [0244] S203 ', by the query instruction matches the pieces of knowledge in the knowledge base dependency, the subtraction of the first score, needs to obtain the overall score of the knowledge base; In a preferred embodiment of the present invention , if the relationship of knowledge fragments belong to the relationship, the points; if knowledge fragments belong to the non-relationship, it is less points;

[0245] S204'、对所述查询指令在表达模板层面上进行打分,得到表达模板得分; [0245] S204 ', the query instructions for scoring the expression level of the template, the template score was expressed;

[0246] S205'、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分; [0246] S205 ', the overall demand knowledge score and the score of the template expression intensity score as a weighted sum query instruction needs;

[0247] S206'、判断所述查询指令需求强度得分是否大于一设定阈值; [0247] S206 ', the query instruction is determined whether the intensity of the demand score greater than a predetermined threshold value;

[0248] S207'、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图; [0248] S207 ', if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intention;

[0249] S208'、若小于所述设定阈值,则判断所述查询指令无明显的需求意图。 [0249] S208 ', if less than the set threshold value, it is determined that the query instruction needs no intent.

[0250] 如图13所示,在本发明第三实施方式、第四实施方式的搜索方法中,所述表达模板库的构建方法,包括以下流程: [0250] As shown, the search method of the third embodiment of the present invention, in the fourth embodiment, the expression construct libraries of template-13, comprising the following processes:

[0251] S300、抽取在用户历史行为库中包含知识片段的查询指令;如在“马六如何”、“斯柯达如何”、“速腾如何”的查询指令中,抽取出知识片段:“马六”、“斯柯达”、“速腾”; [0251] S300, to extract the user's historical behavior database query instruction included knowledge fragments; as in "six horses how to", "Skoda how to", "Sagitar how to" query command, extract the pieces of knowledge: "Six horses" "Skoda", "suteng";

[0252] S301、将所述知识库片段替换成通用符号,生成候选表达模板;例如:“【汽车品牌/型号】”符号,即生成“【汽车品牌/型号】如何”的候选表达模板; [0252] S301, the knowledge base into a common symbol replaced fragment, the expression of generating candidate template; for example: "[brands / models]" symbol, i.e., generated "[brands / models of how]" Expression candidate template;

[0253] S302、统计生成的候选表达模板符合的知识库片段的数量; [0253] S302, the number of segments of the knowledge base to generate statistical template matching candidate expression;

[0254] S303、判断所述数量是否大于设定的阈值; [0254] S303, determines whether the number is greater than the set threshold;

[0255] S304、若大于设定的阈值,则将所述候选表达模板作为表达模板,并存于数据库中,生成表达模板库; [0255] S304, if more than a set threshold, then the candidate templates the expression as an expression of the template, exist in the database, generating an expression template library;

[0256] S305、若小于设定阈值,则舍弃所述候选表达模板。 [0256] S305, if less than a set threshold value, the expression of the candidate template is discarded.

[0257] 通过上述的搜索方法及搜索引擎,本发明一种实施方式的在线界面如图14所示,用于在浏览器中打开本发明搜索引擎的在线界面,并在对话框中输入查询指令“手机诺基亚”,通过上述的搜索方法及搜索系统,可判断出该查询指令包括了商品类的需求意图,故在本发明的搜索方法及搜索系统中,可将“手机诺基亚”这个查询指令在商品垂直搜索数据库中进行搜索,同时,插入该垂直搜索结果至网页存储库中搜索的结果中,如图的A部分, 当用户点击所述垂直搜索结果时,即可跳转至新页面中,如图15所示,该新页面中包含了具有商品类需求意图的检索结果,从图中B部分可看出,这条检索结果中并未包括“手机” 这个文本信息,即是通过本发明的语义扩展得到的搜索结果。 Online interface [0257] By the above-described search method and a search engine, an embodiment of the present invention is shown in Figure 14, the search engine of the present invention for opening a browser-line interface and query instruction input in the dialog box "Nokia", by the above-described search method and search system may determine that the query instruction includes requirements intended commodity, so the search method and search system of the present invention, a "Nokia" query command product vertical search database search, while inserted into the vertical search results to the search results page repository, as shown in part a, when the user clicks the vertical search results, to jump to a new page, As shown, the new page containing search results 15 having a commodity intended requirements, part B can be seen from the figure, the search result in this text does not include the "mobile phone" information, that is, through the present invention, search results obtained semantic expansion.

[0258] 综上所述可知,本发明在知识库的基础上,更好的理解用户输入的查询指令,分析查询指令的以图,解析查询指令的结构,对查询指令进行语义内容扩充,从而更好的指导搜索引擎选择优质的资源满足用户的搜索需求,使得用户搜索效率提高,节约网络流量。 [0258] In summary above, the present invention is based on the knowledge base, a better understanding of query instruction input by a user, an instruction to analyze the query FIG resolution structure of the query instruction, the query expansion instruction semantic content, thereby better guide the search engine to select high-quality resources to meet the needs of the user's search, allowing users to search more efficient, saving network traffic.

[0259] 应当理解,虽然本说明书按照实施方式加以描述,但并非每个实施方式仅包含一个独立的技术方案,说明书的这种叙述方式仅仅是为清楚起见,本领域技术人员应当将说明书作为一个整体,各实施例中的技术方案也可以经适当组合,形成本领域技术人员可以理解的其他实施方式。 [0259] It should be understood that while the present specification be described in terms of embodiments, but not every embodiment contains only a separate aspect, this narrative description only for the sake of clarity, those skilled in the art should be used as a specification overall, the technical solutions in the respective embodiments may be suitably combined to form other embodiments of the present art can be appreciated in the art.

[0260] 上文所列出的一系列的详细说明仅仅是针对本发明的可行性实施方式的具体说明,它们并非用以限制本发明的保护范围,凡未脱离本发明技艺精神所作的等效实施方式或变更均应包含在本发明的保护范围之内。 [0260] A series of the detailed description set forth above is merely for the feasibility of specifically described embodiments of the present invention, they are not intended to limit the scope of the present invention, who have not departing from the spirit of the present invention the equivalent skills made embodiment or changes be included within the scope of the present invention.

Claims (28)

1. 一种搜索方法,其特征在于,所述搜索方法包括以下步骤:S1、接收查询指令;S2、基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图;S3、将带有需求意图的所述查询指令在数据库中搜索,得到搜索结果;S4、输出所述搜索结果。 A search method, characterized in that said search method comprising the steps of: S1, receiving a query command; S2, based on knowledge of the intended query instruction needs analysis, clear the query instruction needs intention; S3 , the query command with a database search intent needs to obtain search results; S4, outputs the search result.
2.根据权利要求1所述的搜索方法,其特征在于,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 2. A search method according to claim 1, wherein said database is a web database repository or the vertical search intention corresponding demand.
3.根据权利要求1所述的搜索方法,其特征在于,在所述S2步骤和S3步骤间,还包括语义扩充步骤:基于所述知识库对所述查询指令进行语义扩充。 3. The search method according to claim 1, characterized in that, between step S2 and the step S3, further comprising the step of semantic expansion: the semantic query expansion instruction based on the knowledge.
4.根据权利要求1所述的搜索方法,其特征在于,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”具体包括以下流程:S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;S201、将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段;S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;S204、判断所述知识库整体需求得分是否大于一设定阈值;S205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图;S206、若小于所述设定阈值, The search method according to claim 1, wherein the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes: S200, user library historical behavior needs to each respective pieces of knowledge in the knowledge base intention score, the respective pieces of knowledge has a corresponding demand intention score; S201, the query command with matching pieces of knowledge, to give matches the query command a fragment of at least knowledge; S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction; S203, by the query instruction matches the pieces of knowledge in the knowledge base dependency, addition and subtraction of said first fraction, to obtain the overall score needs knowledge; S204, determines the overall demand knowledge score is greater than a set threshold; S205, if greater than the set threshold value places the knowledge library overall demand type needs as the highest score of the query command needs intent; S206, if less than the set threshold value, 则判断所述查询指令无明显需求意图。 Determining that the query instruction needs no intent.
5.根据权利要求1所述的搜索方法,其特征在于,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”具体包括以下流程:S200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;S201、将所述查询指令与知识片段和表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一表达模板;S202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;S203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;S204、对所述查询指令在表达模板层面上进行打分,得到表达模板得分;S205、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分;S206、判断 The search method according to claim 1, wherein the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes: S200, user historical behavior to the individual needs of each library fragment knowledge of the intention of scoring knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring; S201, the query instructions and pieces of knowledge and expression template matching, get the query command and a knowledge of at least a fragment of the expression matches the template; S202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction; S203, pieces of knowledge through the query command matches in dependency of the knowledge base, the first subtraction score, needs to obtain the overall score knowledge; S204, instruction in the query expression level is scored on the template, resulting in the expression template score; S205, knowledge base expression template overall demand scores score as a weighted sum query instruction needs intensity score; S206, determination 所述查询指令需求强度得分是否大于一设定阈值;S207、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图;S208、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 The query instruction needs whether the intensity score greater than a predetermined threshold value; S207, if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intent; S208, if less than the setting a threshold value, it is determined that the query instruction needs no intent.
6. 一种搜索方法,其特征在于,所述搜索方法包括以下步骤:51、接收查询指令;52、基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图,同时,基于所述知识库对所述查询指令进行语义扩充;53、将带有需求意图并扩充语义的查询指令在数据库中搜索,得到搜索结果;54、输出所述搜索结果。 A search method, characterized in that said search method comprising the steps of: 51, receiving a query command; 52, based on knowledge of the intended query instruction needs analysis, needs a clear intention of the query command, at the same time based on the knowledge of the semantic query expansion instruction; 54, outputting the search result; 53, with the expansion of the semantic intent and requirements of the database search query instruction, obtain search results.
7.根据权利要求6所述的搜索方法,其特征在于,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 7. A method of searching according to claim 6, wherein said database is a web page or store the corresponding demand intended vertical search database.
8.根据权利要求6所述的搜索方法,其特征在于,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”具体包括以下流程:5200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;5201、将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段;5202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;5203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;5204、判断所述知识库整体需求得分是否大于一设定阈值;5205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图;5206、若小于所述设定阈值, 8. The search method according to claim 6, characterized in that the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes: 5200, by the user library historical behavior needs to each respective pieces of knowledge in the knowledge base intention score, the respective pieces of knowledge has a corresponding demand intention score; 5201, the query command and knowledge fragment matches the query command to obtain match a fragment of at least knowledge; 5202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction; 5203 by the query instruction matches the pieces of knowledge in the knowledge base dependency, addition and subtraction of said first fraction, to obtain the overall score needs knowledge; 5204, it is determined whether the overall demand knowledge score is greater than a set threshold; 5205, if greater than the set threshold value, places the knowledge library overall demand type needs as the highest score of the query instructions intended to demand; 5206, if less than the set threshold value, 则判断所述查询指令无明显需求意图。 Determining that the query instruction needs no intent.
9.根据权利要求6所述的搜索方法,其特征在于,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”具体包括以下流程:5200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;5201、将所述查询指令与知识片段和表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一表达模板;5202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;5203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;5204、对所述查询指令在表达模板层面上进行打分,得到表达模板得分;5205、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分;5206、判断 9. The search method according to claim 6, characterized in that the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes: 5200, by the user historical behavior to the individual needs of each library fragment knowledge of the intention of scoring knowledge Base so that each segment has a corresponding knowledge of the needs of the intention of scoring; 5201, the query instructions and pieces of knowledge and expression template matching, get the query command and a knowledge of at least a fragment of the template match expression; 5202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction; 5203, through knowledge of the segments that match the query instructions in dependency of the knowledge base, the first subtraction score, needs to obtain the overall score knowledge; 5204, instructions in the query expression level is scored on the template, resulting in the expression template score; 5205, knowledge base expression template overall demand scores score as a weighted sum query instruction needs intensity score; 5206, Analyzing 所述查询指令需求强度得分是否大于一设定阈值;5207、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图;5208、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 The query instruction needs whether the intensity score greater than a predetermined threshold value; 5207, if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intent; 5208, if less than the setting a threshold value, it is determined that the query instruction needs no intent.
10. 一种搜索方法,其特征在于,所述搜索方法包括以下步骤:Si、接收查询指令;.52、基于知识库和表达模板库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图;.53、将带有需求意图的所述查询指令在数据库中搜索,得到搜索结果;.54、输出所述搜索结果。 10. A search method, characterized in that said search method comprising the steps of: Si, receiving a query command; .52, based on knowledge, and the query expression template library instruction needs intent analysis, clear the query instruction demand intent; .53, with a query instruction to search the database requirements intended to obtain search results; .54, the output of the search result.
11.根据权利要求10所述的搜索方法,其特征在于,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 11. The search method according to claim 10, wherein said database is a web database repository or the vertical search intention corresponding demand.
12.根据权利要求10所述的搜索方法,其特征在于,在所述S2步骤和S3步骤间,还包括语义扩充步骤:基于所述知识库对所述查询指令进行语义扩充。 12. The search method according to claim 10, characterized in that, between step S2 and the step S3, further comprising the step of semantic expansion: the semantic query expansion instruction based on the knowledge.
13.根据权利要求10所述的搜索方法,其特征在于,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”具体包括以下流程:.5200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;.5201、将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段;.5202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;.5203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;.5204、判断所述知识库整体需求得分是否大于一设定阈值;.5205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图;.5206、若小于所述设定 13. The search method according to claim 10, wherein the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes: .5200, by user behavior history database to the individual needs of each of the pieces of knowledge in the knowledge base intention score, the respective pieces of knowledge has a corresponding demand intention score; .5201, the query command and knowledge fragment matches the query command to obtain phase matching at least a fragment of knowledge; .5202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction; .5203 ​​by the query instruction matches the pieces of knowledge in affiliation knowledge base, the first subtraction score, needs to obtain the overall score knowledge; .5204, determining the overall demand knowledge score is greater than a set threshold; .5205, if greater than the set threshold value , places the highest scoring overall demand knowledge of the needs of the type of query instruction needs as intended; .5206, if less than the set 阈值,则判断所述查询指令无明显需求意图。 The threshold value, it is determined that the query instruction needs no intent.
14.根据权利要求10所述的搜索方法,其特征在于,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”具体包括以下流程:.5200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;.5201、将所述查询指令与知识片段和表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一表达模板;.5202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;.5203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;.5204、对所述查询指令在表达模板层面上进行打分,得到表达模板得分;.5205、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分;.5206 14. The search method according to claim 10, wherein the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes: .5200, by user behavior history database knowledge to the individual needs of each segment of the knowledge base of intent scoring, so that each segment has a corresponding knowledge of the needs of the intention of scoring; .5201, the query instructions and pieces of knowledge and expression template matching, get the and at least a fragment of a knowledge expression template match query instruction; .5202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction; .5203, with the query instructions by affiliation matching pieces of knowledge in the knowledge base, the first subtraction score, needs to obtain the overall score knowledge; .5204, instruction in the query expression level is scored on the template, the template score was expressed; .5205, the overall demand knowledge score and the score of the template expression intensity score as a weighted sum query instruction needs; .5206 、判断所述查询指令需求强度得分是否大于一设定阈值;.5207、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图;.5208、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 Determining whether the intensity of the query command needs score greater than a predetermined threshold value; .5207, if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intent; .5208, If less than the set threshold value, it is determined that the query instruction needs no intent.
15.根据权利要求10所述的搜索方法,其特征在于,所述表达模板库的构建方法,包括以下流程:·5300、抽取在用户历史行为库中包含知识片段的查询指令;·5301、将所述知识库片段替换成通用符号,生成候选表达模板;·5302、统计生成的所述候选表达模板符合的知识库片段的数量;·5303、判断所述数量是否大于设定阈值;·5304、若大于设定阈值,则将所述候选表达模板作为表达模板,并存于数据库中,生成表达模板库;·5305、若小于设定阈值,则舍弃所述候选表达模板。 15. The search method according to claim 10, wherein the expression construct template library method, comprising the following processes: · 5300, extracting query instruction fragments contained in the knowledge database user behavior history; * 5301, the said knowledge base into a common symbol replaced fragment, the expression of generating candidate template; * number 5302, the statistics generation segment of the knowledge base expressed candidate template matching; * 5303, determines whether the number is greater than the set threshold; - 5304, If the value is greater than the threshold, then the candidate templates the expression as an expression of the template, it exists in the database, generating an expression library template; * 5305, if less than a set threshold value, the candidate is discarded expression template.
16. 一种搜索方法,其特征在于,所述搜索方法包括以下步骤:·51、接收查询指令;·52、基于知识库和表达模板库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图,同时,基于所述知识库对接收到的查询指令进行语义扩充;·53、将带有需求意图并扩充语义的查询指令在数据库中搜索,得到搜索结果;·54、输出所述搜索结果。 16. A search method, characterized in that said search method comprising the following steps: 51 receiving the query instruction; * 52, based on knowledge, and the query expression template library instruction needs intent analysis, clear the query demand command is intended, at the same time, expansion of semantic knowledge base based on the query instruction received; * 53, with the expansion of the semantic intent and requirements of the database search query instruction, obtain search results; * 54, the output above search results.
17.根据权利要求16所述的搜索方法,其特征在于,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 17. The search method according to claim 16, wherein said database is a web database repository or the vertical search intention corresponding demand.
18.根据权利要求16所述的搜索方法,其特征在于,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”具体包括以下流程:·5200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;·5201、将所述查询指令与知识片段匹配,得到与所述查询指令相匹配的至少一个知识片段;·5202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;·5203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;·5204、判断所述知识库整体需求得分是否大于一设定阈值;·5205、若大于所述设定阈值,则以所述知识库整体需求得分最高的需求类型作为所述查询指令的需求意图;·5206、若小于所 18. The search method according to claim 16, wherein the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes: · 5200, by user behavior history database knowledge to the individual needs of each segment of the knowledge base of intent scoring, so that each segment has a corresponding knowledge of the needs of the intention of scoring; · 5201, the query instruction and knowledge fragment matched to give instructions to the inquiry phase matching at least a fragment of knowledge; - 5202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first fraction; - 5203, by the instruction matches the query in the knowledge fragment affiliation knowledge base, the first subtraction score, needs to obtain the overall score of the knowledge base; * 5204, determines whether the overall demand knowledge score greater than a predetermined threshold value; * 5205, if greater than the set threshold value , places the highest scoring overall demand knowledge of the needs of the type of query instruction needs as intended; * 5206, if less than the 设定阈值,则判断所述查询指令无明显需求意图。 Setting a threshold value, it is determined that the query instruction needs no intent.
19.根据权利要求16所述的搜索方法,其特征在于,所述“基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图”具体包括以下流程:·5200、通过用户历史行为库给知识库中的各个知识片段的各个需求意图打分,使各个知识片段都具有相应的需求意图得分;·5201、将所述查询指令与知识片段和表达模板匹配,得到与所述查询指令相匹配的至少一个知识片段和一表达模板;·5202、将与所述查询指令相匹配的知识片段的需求意图得分加总,得到第一分数;·5203、通过与所述查询指令相匹配的知识片段在所述知识库中的从属关系,加减所述第一分数,得到知识库整体需求得分;·5204、对所述查询指令在表达模板层面上进行打分,得到表达模板得分;·5205、将知识库整体需求得分与表达模板得分的加权之和作为查询指令需求强度得分 19. The search method according to claim 16, wherein the "Knowledge-based instruction needs the query intent analysis, clear the query instruction needs intention" specifically includes the following processes: · 5200, by user behavior history database knowledge to the individual needs of each segment of the knowledge base of intent scoring, so that each segment has a corresponding knowledge of the needs of the intention of scoring; · 5201, the query instructions and pieces of knowledge and expression template matching, get the and at least a fragment of a knowledge expression template match query instruction; * 5202, the demand for the query instructions intended to match the pieces of knowledge score summed to obtain a first score; * 5203, through the query command with matching pieces of knowledge affiliation in the knowledge Base, plus or minus the first score to give the overall demand score knowledge base; · 5204, the query expression is scored on instruction in the template level, resulting in the expression templates score; · 5205, the overall demand for knowledge and expression templates score score as the weighted sum of the query instructions strength rating requirements ·5206、判断所述查询指令需求强度得分是否大于一设定阈值;·5207、若大于所述设定阈值,则以查询指令需求强度得分最高的需求类型作为所述查询指令的需求意图;·5208、若小于所述设定阈值,则判断所述查询指令无明显需求意图。 * 5206, determines whether the query instruction needs intensity score greater than a predetermined threshold value; * 5207, if greater than the set threshold value, the strength of demand query instruction places the highest score of the demand type as a query instruction needs intent; - 5208, if less than the set threshold value, it is determined that the query instruction needs no intent.
20.根据权利要求16所述的搜索方法,其特征在于,所述表达模板库的构建方法,包括以下流程:5300、抽取在用户历史行为库中包含知识片段的查询指令;5301、将所述知识库片段替换成通用符号,生成候选表达模板;5302、统计生成的所述候选表达模板符合的知识库片段的数量;5303、判断所述数量是否大于设定阈值;5304、若大于设定阈值,则将所述候选表达模板作为表达模板,并存于数据库中,生成表达模板库;5305、若小于设定阈值,则舍弃所述候选表达模板。 20. The search method according to claim 16, wherein the expression construct template library method, comprising the following processes: 5300, extracts the user behavior history information database query instruction fragment comprising; 5301, the Alternatively the knowledge base segments into common symbol, generating candidate expression template; 5302, the number of statistics generated knowledge base segment of the expression of the candidate template matching; 5303, determines whether the number is greater than the set threshold; 5304, if greater than the set threshold value , then the expression of the candidate template as a template expression, co-exist in the database, generating an expression library template; 5305, if less than a set threshold value, the candidate is discarded expression template.
21. 一种搜索引擎,其特征在于,所述搜索引擎包括:UI模块,用于接收查询指令,且所述UI模块还用于接收搜索模块返回的搜索结果,并将所述搜索结果拼装为结果页面后输出;需求意图分析模块,用于基于知识库对所述查询指令进行需求意图分析,明确所述查询指令的需求意图;搜索模块,用于将带有需求意图的所述查询指令在数据库中搜索,得到搜索结果; 知识库,用于存储先验知识。 21. A search engine, wherein the search engine comprises: UI module, configured to receive a query instruction, and the UI module is further configured to receive the search module returns the search results, and the search result for the assembly results page output; intention demand analysis module, based on knowledge of the intention of the query instruction needs analysis, a clear intention of the query instruction needs; a search module for the query instruction with intent in demand Search the database to obtain search results; knowledge base for storing a priori knowledge.
22.根据权利要求21所述的搜索引擎,其特征在于,所述搜索引擎还包括:web服务模块,用于通过网络协议接收客户端发出的查询指令,并将所述查询指令转到所述UI模块,且所述web服务模块还用于接收所述UI模块返回的结果页面,并将所述结果页面返回至所述客户端。 22. The search engine of claim 21, wherein said search engine further comprising: web service module for receiving a query instruction sent by the client through the network protocol, and the query instruction to the UI module and the web service module is further configured to receive the UI module returns the result page, and the result page returned to the client.
23.根据权利要求21所述的搜索引擎,其特征在于,所述搜索引擎还包括: 用户历史行为库,用于存储用户历史搜索记录。 23. The search engine of claim 21, wherein said search engine further comprising: a user behavior history database for storing user history search history.
24.根据权利要求23所述的搜索引擎,其特征在于,所述用户历史搜索记录包括:查询指令、查询次数,以及加权点击数。 24. The search engine of claim 23, wherein the user search history record comprising: a query instructions, queries, and the weighting clicks.
25.根据权利要求23或24所述的搜索引擎,其特征在于,所述搜索引擎还包括:表达模板挖掘模块,用于根据所述知识库中的知识片段和所述用户历史行为库中的用户历史查询指令,挖掘表达模板,并将所述表达模板存储于表达模板库; 表达模板库,用于存储由所述表达模板挖掘模块挖掘出的表达模板。 25. The search engine of claim 23 or claim 24, wherein said search engine further comprises: a template expressing mining module, according to the knowledge in the knowledge base of the segment and the user behavior history library user history query instruction, mining expression template, and the template stored in the expression of the expression template library; expression template library, the template for the expression of the expression template storage mining module excavated.
26.根据权利要求21所述的搜索引擎,其特征在于,所述搜索引擎还包括: 结构分类模块,用于基于所述知识库对所述查询指令进行语义扩充。 26. The search engine of claim 21, wherein said search engine further comprising: a structural classification module based on the knowledge of the semantic query expansion instruction.
27.根据权利要求21所述的搜索引擎,其特征在于,所述数据库为网页存储库或与所述需求意图相对应的垂直搜索数据库。 27. The search engine of claim 21, wherein said database is a web database repository or the vertical search intention corresponding demand.
28.根据权利要求27所述的搜索引擎,其特征在于,所述网页存储库用于存储网页数据和该网页数据的索引信息;所述垂直搜索数据库用于存储特定类别数据和该特定类别数据的索引信息。 28. A search engine as recited in claim 27, wherein the web page repository for storing index information of the web page data and page data; vertical search the database for storing data of a particular category and the category-specific data the index information.
CN 201110038433 2011-02-15 2011-02-15 Search method and search engine CN102096717B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201110038433 CN102096717B (en) 2011-02-15 2011-02-15 Search method and search engine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201110038433 CN102096717B (en) 2011-02-15 2011-02-15 Search method and search engine

Publications (2)

Publication Number Publication Date
CN102096717A true true CN102096717A (en) 2011-06-15
CN102096717B CN102096717B (en) 2013-01-16

Family

ID=44129812

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201110038433 CN102096717B (en) 2011-02-15 2011-02-15 Search method and search engine

Country Status (1)

Country Link
CN (1) CN102096717B (en)

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102402619A (en) * 2011-12-23 2012-04-04 广东威创视讯科技股份有限公司 Search method and device
CN102591932A (en) * 2011-12-23 2012-07-18 优视科技有限公司 Voice search method, voice search system, mobile terminal and transfer server
CN102982025A (en) * 2011-09-02 2013-03-20 北京百度网讯科技有限公司 Identification method and device for searching requirement
CN103136221A (en) * 2011-11-24 2013-06-05 北京百度网讯科技有限公司 Method capable of generating requirement template and requirement identification method and device
CN103186573A (en) * 2011-12-29 2013-07-03 北京百度网讯科技有限公司 Method for determining search requirement strength, requirement recognition method and requirement recognition device
CN103389988A (en) * 2012-05-10 2013-11-13 腾讯科技(深圳)有限公司 Method and device for guiding user to carry out information search
CN103530385A (en) * 2013-10-18 2014-01-22 北京奇虎科技有限公司 Method and device for searching for information based on vertical searching channels
CN103559253A (en) * 2013-10-31 2014-02-05 北京奇虎科技有限公司 Related vertical resource search method and equipment
CN103577560A (en) * 2013-10-24 2014-02-12 华为技术有限公司 Method and device for inputting data base operating instructions
CN104252298A (en) * 2013-06-25 2014-12-31 刘建 Information management system based on external equipment of electronic device
CN104298658A (en) * 2014-10-29 2015-01-21 百度在线网络技术(北京)有限公司 Method and device for acquiring search result
CN104424216A (en) * 2013-08-23 2015-03-18 佳能株式会社 Method and device for intention digging
CN104598617A (en) * 2015-01-30 2015-05-06 百度在线网络技术(北京)有限公司 Method and device for displaying search results
CN102999520B (en) * 2011-09-15 2016-04-27 北京百度网讯科技有限公司 A method and apparatus searches identified needs

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080104061A1 (en) * 2006-10-27 2008-05-01 Netseer, Inc. Methods and apparatus for matching relevant content to user intention
CN101398810A (en) * 2007-09-30 2009-04-01 日电(中国)有限公司 Self-adapting service choice device and method thereof, enquiry system and method thereof
CN101751422A (en) * 2008-12-08 2010-06-23 北京摩软科技有限公司 Method, mobile terminal and server for carrying out intelligent search at mobile terminal

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080104061A1 (en) * 2006-10-27 2008-05-01 Netseer, Inc. Methods and apparatus for matching relevant content to user intention
CN101398810A (en) * 2007-09-30 2009-04-01 日电(中国)有限公司 Self-adapting service choice device and method thereof, enquiry system and method thereof
CN101751422A (en) * 2008-12-08 2010-06-23 北京摩软科技有限公司 Method, mobile terminal and server for carrying out intelligent search at mobile terminal

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102982025B (en) * 2011-09-02 2016-05-11 北京百度网讯科技有限公司 A method and apparatus for identifying search needs
CN102982025A (en) * 2011-09-02 2013-03-20 北京百度网讯科技有限公司 Identification method and device for searching requirement
CN102999520B (en) * 2011-09-15 2016-04-27 北京百度网讯科技有限公司 A method and apparatus searches identified needs
CN103136221A (en) * 2011-11-24 2013-06-05 北京百度网讯科技有限公司 Method capable of generating requirement template and requirement identification method and device
CN102402619B (en) 2011-12-23 2013-10-30 广东威创视讯科技股份有限公司 Search method and device
CN102402619A (en) * 2011-12-23 2012-04-04 广东威创视讯科技股份有限公司 Search method and device
CN102591932A (en) * 2011-12-23 2012-07-18 优视科技有限公司 Voice search method, voice search system, mobile terminal and transfer server
CN103186573A (en) * 2011-12-29 2013-07-03 北京百度网讯科技有限公司 Method for determining search requirement strength, requirement recognition method and requirement recognition device
CN103186573B (en) * 2011-12-29 2016-05-18 北京百度网讯科技有限公司 A method of determining a search strength requirements, identify the need for a method and apparatus
CN103389988A (en) * 2012-05-10 2013-11-13 腾讯科技(深圳)有限公司 Method and device for guiding user to carry out information search
CN104252298A (en) * 2013-06-25 2014-12-31 刘建 Information management system based on external equipment of electronic device
CN104424216A (en) * 2013-08-23 2015-03-18 佳能株式会社 Method and device for intention digging
CN103530385A (en) * 2013-10-18 2014-01-22 北京奇虎科技有限公司 Method and device for searching for information based on vertical searching channels
CN103577560A (en) * 2013-10-24 2014-02-12 华为技术有限公司 Method and device for inputting data base operating instructions
CN103559253A (en) * 2013-10-31 2014-02-05 北京奇虎科技有限公司 Related vertical resource search method and equipment
CN104298658A (en) * 2014-10-29 2015-01-21 百度在线网络技术(北京)有限公司 Method and device for acquiring search result
CN104598617A (en) * 2015-01-30 2015-05-06 百度在线网络技术(北京)有限公司 Method and device for displaying search results

Also Published As

Publication number Publication date Type
CN102096717B (en) 2013-01-16 grant

Similar Documents

Publication Publication Date Title
US7634466B2 (en) Realtime indexing and search in large, rapidly changing document collections
US20090006343A1 (en) Machine assisted query formulation
US20090248661A1 (en) Identifying relevant information sources from user activity
US20070192300A1 (en) Method and system for determining relevant sources, querying and merging results from multiple content sources
US20060173834A1 (en) Table querying
US20120158685A1 (en) Modeling Intent and Ranking Search Results Using Activity-based Context
US20090006345A1 (en) Voice-based search processing
US20060271520A1 (en) Content-based implicit search query
US8024327B2 (en) System and method for measuring the quality of document sets
US8504437B1 (en) Dynamically selecting and presenting content relevant to user input
US20120117059A1 (en) Ranking Authors in Social Media Systems
US20070136256A1 (en) Method and apparatus for representing text using search engine, document collection, and hierarchal taxonomy
US20120143880A1 (en) Methods and apparatus for providing information of interest to one or more users
US20080109285A1 (en) Techniques for determining relevant advertisements in response to queries
US20070043759A1 (en) Method for data management and data rendering for disparate data types
US20100241647A1 (en) Context-Aware Query Recommendations
US20120203734A1 (en) Automatic mapping of a location identifier pattern of an object to a semantic type using object metadata
US7289985B2 (en) Enhanced document retrieval
US20120059838A1 (en) Providing entity-specific content in response to a search query
US20090240674A1 (en) Search Engine Optimization
US20100185691A1 (en) Scalable semi-structured named entity detection
US8458165B2 (en) System and method for applying ranking SVM in query relaxation
US20080270380A1 (en) Method for Determining Contextual Summary Information Across Documents
US20120054440A1 (en) Systems and methods for providing a hierarchy of cache layers of different types for intext advertising
US20050234952A1 (en) Content propagation for enhanced document retrieval

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted