【正文】 41。查詢結(jié)果的獨(dú)特類設(shè)置介紹不但極大地減少了反復(fù)查詢多個(gè)數(shù)據(jù)庫(kù)所付出的時(shí)間和費(fèi)用，并且可以做到高效率檢索相關(guān)的查詢結(jié)果?；诜诸愒~典和嵌入搜索策略的信息庫(kù)，允許用戶使用“自然”語(yǔ)言來(lái)進(jìn)行專業(yè)的信息查詢。當(dāng)一名用戶從類型2執(zhí)行一次查詢時(shí)，系統(tǒng)提出查詢結(jié)果自動(dòng)地運(yùn)用集合2類別對(duì)用戶。例如，一套類別為用戶的第一個(gè)典型類型建立，并且不同的套類別也許為用戶的第二個(gè)典型類型建立。例如，一本典型的商業(yè)雜志也許包含信息的幾個(gè)類型，例如社論、規(guī)則專欄、特寫(xiě)、新聞、產(chǎn)品公告和事件日歷。用戶發(fā)出一個(gè)命令然后請(qǐng)求，例如“VI CO 1”，查驗(yàn)從名單挑選的完全文件，在這種情況下，給關(guān)于身分專家的完全信息和證件。如圖2，查詢生成過(guò)程包括分類詞詞典和筆記的一個(gè)知識(shí)庫(kù)和運(yùn)用嵌入被定定義的復(fù)雜戰(zhàn)略。4. 圖例簡(jiǎn)要說(shuō)明圖1是信息查詢系統(tǒng)總流程圖；圖2是系統(tǒng)制定查詢和搜索過(guò)程圖。然后提供給相對(duì)于用戶相關(guān)查詢的結(jié)果與在該查詢結(jié)果中的每個(gè)類別相關(guān)文檔數(shù)量的統(tǒng)計(jì)。只要每次查詢到一個(gè)原文歸檔數(shù)據(jù)的固有部分，所有建立的文檔就能返回到其組織過(guò)程。查詢創(chuàng)建過(guò)程包含一個(gè)知識(shí)庫(kù)，該知識(shí)庫(kù)包括被預(yù)先確定和嵌入復(fù)雜查詢的分類詞典，或者是自然語(yǔ)言的處理，或者模糊邏輯，或者樹(shù)型結(jié)構(gòu)，或者等級(jí)關(guān)系，或者是一套尋求信息的公式化查詢命令。該系統(tǒng)不僅包括存儲(chǔ)廣泛數(shù)據(jù)領(lǐng)域的復(fù)合數(shù)據(jù)源記錄，還包括多個(gè)文件類型的某些原始記錄。然而，當(dāng)面對(duì)變得更大來(lái)源分組或需要更加全面的查詢結(jié)果時(shí)，這個(gè)問(wèn)題就更加明顯，人們尋找的信息經(jīng)常面對(duì)大量未組織的結(jié)果集合，這樣就需要增加過(guò)濾查詢的重要任務(wù)。即使你按照這樣操作，也有可能錯(cuò)過(guò)相關(guān)的答案，因?yàn)樗锌赡芟嚓P(guān)的數(shù)據(jù)庫(kù)或信息來(lái)源并不在每一次搜索查詢中。而目前的系統(tǒng)可以搜尋大的數(shù)據(jù)，在這過(guò)程中要求人們尋求信息或試圖修改他們的查詢條件，以減少不必要的搜索結(jié)果（消滅潛在的相關(guān)結(jié)果），使用戶查詢到真正要查的數(shù)據(jù)。在查詢過(guò)程中，會(huì)不斷的重復(fù)查詢每一個(gè)數(shù)據(jù)來(lái)源或一組數(shù)據(jù)源，為了確保搜索出所有相關(guān)的文件，這個(gè)重復(fù)是非常必要的。今天所使用的多數(shù)系統(tǒng)實(shí)際上采用的是同一方式。關(guān)鍵詞：信息管理；檢索系統(tǒng)；面向?qū)ο?. 簡(jiǎn)介信息的存儲(chǔ)，查詢和檢索系統(tǒng),主要應(yīng)用原文檔數(shù)據(jù)比較大的文檔，利用搜索條件和索引字段可以快速查詢結(jié)果。該系統(tǒng)包括多個(gè)查詢產(chǎn)生過(guò)程和一個(gè)搜索過(guò)程。 FIG. 3 is a diagram illustrating a sorting process for organizing and presenting search results. MODE FOR CARRYING OUT THE INVENTION As is illustrated in the block diagram of FIG. 1 , the information retrieval system of the invention includes an input/output process ,a query generation process, a search process that involves a large domain of textual data (typically in the multiple gigabyte range), an organizing process, presentation of the information to the user, and a process to identify and characterize the types of documents contained in the large domain of data.Turning now to FIG. 2, the query generation process preferably includes a knowledge base containing a thesaurus and a note pad, and preferably utilizes embedded predefined plex Boolean strategies. Such a system allows the user to enter their description of the information needed using simple words/phrases made up of natural language and to rely on the system to assist in generating the full search query, which would include, ., synonyms and alternate phraseology. The user can then request, by a mand such as VI CO 1, to view the plete document selected from the list, giving, in this case, plete information about the identity and credentials of the expert.FIG. 3 illustrates how five typical sources of information (., source records) can be sorted into many document types and then subsequently into categories. For example, a typical trade magazine may contain several types of information such as editorials, regular columns, feature articles, news, product announcements, and a calendar of events. Thus, the trade magazine (., the source record) may be sorted into these various document types, and these document types in turn may be categorized or grouped into categories contained in one or more sets of categories。 and means for categorizing documents responsive to the query based on document type, including means for generating a summary of the number of documents responsive to the query which fall within various predetermined categories of document types. The query generation process may contain a knowledge base including a thesaurus that has predetermined and embedded plex search queries, or use natural language processing, or fuzzy logic, or tree structures, or hierarchical relationship or a set of mands that allow persons seeking information to formulate their queries. The search process can utilize any index and search engine techniques including Boolean, vector, and probabilistic as long as a substantial portion of the entire domain of archived textual data is searched for each query and all documents found are returned to the organizing process. The sorting/categorization process prepares the search results for presentation by assembling the various document types retrieved by the search engine and then arranging these basic document types into sometimes broader categories that are readily understood by and relevant to the search results are then presented to the user and arranged by category along with an indication as to the number of relevant documents found in each category. The user may then examine search results in multiple formats, allowing the user to view as much of the document as the user deems necessary. DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram illustrating an information retrieval system of the invention。 that is, users log on (through a puter terminal or personal microputer, and typically from a remote location), select a source of information (., a particular database) which is usually something less than the plete domain, formulate a query, launch the search, and then review the search results displayed on the terminal or microputer, typically with documents (or summaries of documents) displayed in reverse chronological order. This process must be repeated each time another source (database) or group of sources is selected (which is frequently necessary in order to insure all relevant documents have been found).Additionally, this process places on the user the burden of organizing and assimilating the multiple results generated from the launch of t