TWI221992B

TWI221992B - Information search method of patent literature

Info

Publication number: TWI221992B
Application number: TW91137075A
Authority: TW
Inventors: Ji-Hung Liau; Jian-Jung Yuan; Mei-Jiuan Jiang
Original assignee: Inst Information Industry
Priority date: 2002-12-23
Filing date: 2002-12-23
Publication date: 2004-10-11
Also published as: TW200411428A

Abstract

The invention relates to a technique adopted to information search of patent literature. The method of the invention looks for the corresponding patent literature matching with the keyword in computer system. Moreover, the computer system further has a keyword type table, enabling to identify the possibly related technique type by comparison of keyword and a keyword type table. Furthermore, employ the possibly related technique type to eliminate few impossibly relevant patent literatures for more search (e.g. full text retrieval) or after proceeding search, let users know what would be the more possibly related patent literatures with the relevant technology type.

Description

玖、發明說明 (發明說明應敘明:發明所屬之技術領· 、先前技術、內容、實施方式及圖式簡單說明) [ 術、發明所屬之技術領本發明係關於一種應、域】用於專利文獻資料檢索之技【二、先前技術】專利文獻資料檢旁的i 、 τ你系的重要性越來越重要，因為專利 =獻本身就是-項非常重要之技術文件，有助於研發人員開發之參考。另外公司在開發新產品時，最好知道有無侵犯他人專利，以你_ H f & 兮〜以便评估是否進入此市場，所以需進行專利檢索。、目别美國專利商標局網站所提供之專利文獻檢索為非常知名，也常被使用之網站。還有其他國家如曰本，中國，英國，歐洲專利局等也提供好用之專利文獻檢索工具。在這些專利文獻檢索工具中，都有提供『關鍵字』查詢，然而有時候僅以『關鍵字』查詢常會查到自己不需要之專利文獻。所以有些官方網站亦提供較高階之查β旬’言如使用者除輸入『關鍵字』外，亦可限制尋找哪些年之專利文獻，或是限制哪些專利申請人所屬之專利文獻等等。其中對於技術之限制條件最重要，也最有用的是技術分類，譬如最有名也被普遍採用的是J p c (International Patent Classification，國際專利分類）分類。限制『國際專利分類』相當好用，但是有個問題是『國際專利分類』之細分類別相當多，一般使用者根本 1221992 不知道自己想查的技術是屬於哪一類，或是使用者必須發相當多的時間去找出自己想查的技術是屬於哪一類。所以問題在於雖然『國際專利分類』對於查詢專利文獻非常有用，但使用者在查詢專利文獻時卻常常不使用『國際專利分類』。另外由於專利文獻越來越龐大，專利文獻資料量增加速度非常快，如果要維持全文檢索，若能先去除一些不可能相關之專利文獻，則可增加檢索速度。因此以上問通疋有需要去解決的。【三、發明内容】本發明之主要目的係在提供專利文獻資料檢索之方法，並且會進行技術類別之判斷，以便可去除一些不可能相關之專利文獻再進行檢索（如全文檢索），或是進行檢索後，讓使用者知道哪些是較有可能之相關技術類別之專利文獻。為達成上述之目的，本發明專利文獻資料檢索之方法，係用於使用者輸入至少一關鍵字於一電腦系統，該電腦系統包括有專利文獻資料庫，使得電腦系統尋找出與該關鍵字相關之專利文獻，另外電腦系統更有一關鍵字類別總表’該關鍵字類別總表紀錄複數之技術類別，以及每一技術類別對應之複數特徵字。本發明之方法主要包括下列步驟：步驟A :將該關鍵字與一關鍵字類別總表進行比 1221992 步驟B :依照來空存+ 女、、""鄉A由複數之技術類別中分知山子至少一有可鉍研、· 丁刀析出該為候關鍵字至少一有可At s…个反取<钗倂頰別干分柄選技術類別；％屬於之技術類別’在此定義步驟C ··根據兮文獻中進行檢c字由屬於候選技術類別之專利步驟D:顯示找依照上述之方法專利文獻。文獻再進行檢索。…則可去除—些不可能相關之專利根據本發明之$ 步驟A :將該β“—實施例方法包括下列步驟：較； '子與一關鍵字類別總表進行比步驟Β :依照步關鍵字至少一有可&驟Α由複數之技術類別中分析出該選技術類別；成屬於之技術類別，在此定義為候说明 Description of the invention (the description of the invention should state: the technical field to which the invention belongs, a brief description of the prior art, the content, the embodiments, and the drawings) Techniques for Retrieval of Patent Documents [II. Prior Technology] The importance of i and τ next to the inspection of patent documents is becoming more and more important, because patent = contribution itself is a very important technical document, which is helpful to R & D personnel. Development reference. In addition, when the company develops a new product, it is best to know whether it has infringed on the patents of others, so you can evaluate whether to enter this market, so a patent search is required. 2. The patent literature search provided by the U.S. Patent and Trademark Office website is a well-known and often used website. There are also other countries such as Japan, China, the United Kingdom, the European Patent Office, etc. which provide useful search tools for patent documents. In these patent document search tools, “keyword” queries are provided, but sometimes only “keyword” queries often find patent documents that they do not need. Therefore, some official websites also provide higher-level search terms. In addition to entering “keywords”, users can also limit which years of patent documents to search, or which patent applicants belong to which patent documents belong. Among them, the most important and most useful is the technical classification. For example, the most famous and widely used is the J p c (International Patent Classification) classification. Restricting the "International Patent Classification" is quite easy to use, but there is a problem that the "International Patent Classification" has a lot of subdivision categories. The average user does not know what kind of technology the user wants to check, or the user must send the equivalent. More time to find out which type of technology you want to check belongs to. The problem is that although the International Patent Classification is very useful for searching patent documents, users often do not use the International Patent Classification when searching for patent documents. In addition, due to the increasing volume of patent documents, the amount of patent documents increases very quickly. If you want to maintain a full-text search, you can increase the search speed if you can remove some impossible patent documents. Therefore, the above questions need to be resolved. [III. Summary of the Invention] The main purpose of the present invention is to provide a method for searching patent document data, and to judge the technical category, so that some patent documents that are not relevant can be removed and then searched (such as full-text search), or After searching, let users know which patent documents are more likely to be related to the technology category. In order to achieve the above object, the method for retrieving patent document data of the present invention is used for a user to input at least one keyword into a computer system. The computer system includes a patent document database, so that the computer system finds the keywords related In the patent literature, in addition, the computer system also has a general category of keyword categories. The general category of the keyword category records plural technical categories, and plural characteristic words corresponding to each technical category. The method of the present invention mainly includes the following steps: Step A: Compare the keyword with a keyword category master list 1221992 Step B: According to the empty deposit + female, " " township A is divided into plural technical categories At least one of Zhishanzi has bismuth research, and Dingdao has separated out the candidate keywords. At least one has Ats ... one back picking < 钗倂 Do n’t cheek and choose the technology category;% of the technology category 'is defined here Step C: Check the c-word in the document according to the patent belonging to the candidate technology category. Step D: Display the patent document according to the method described above. The literature was searched again. … Then can be removed—some impossible related patents according to the present invention. Step A: The β ”—the embodiment method includes the following steps: Compare; 'Compared with a keyword category summary table Step B: Follow the key At least one word is available & Step A The selected technology category is analyzed from a plurality of technology categories; The technology category to which it belongs is defined here as a candidate

步驟C :根據兮M 步驟D:顯* =鍵字進行專敎獻的檢索；以及有參考步驟⑽找到之專利文獻’其中顯示之方式係 , ^ I之候選技術類別。用者知道：些法，則可在進行檢索後，讓使有可能之相關技術類別之專利文獻。【四、實施方式】特兴：肊·貝審查委員能更瞭解本發明之技術内容，特舉較佳具體實施例說明如下。疒，：七卩關於專利檢索之運用係在-電腦系統上進Step C: According to step M: Step D: Display * = key to perform a dedicated search; and Patent Documents found with reference to Step ’, where the method shown is ^ I candidate technology category. The user knows that these methods can be used to make patent documents of related technical categories possible after a search. [Fourth, the implementation mode] Special promotion: The 贝 · Bei review committee can better understand the technical content of the present invention, and the preferred specific embodiments are described below.疒，：七卩 's application of patent search is based on-computer system

/味於電腦系統為一相當已知之裝置，任何具有-般知識之μ·—誓i I 此灯叢人士都知道，且本發明並非在改變電腦系 8 明更體，因此在此不再贅述電腦系統之功能。在本發 ’電腦系統包括有專利文獻資料庫以及檢索引擎，關電腦系統可以利用使用者輸入之關鍵字尋找出相 —之專利文獻。而本發明特殊之處在於電腦系統内需有關鍵字類別總表，使得本發明之方法才得以運作，此以下先介紹關鍵字類別總表的產生方式，以及其思、義為何。請參見圖1係說明關鍵字類別總表2 0產生方式之流程圖，並請一併參考圖2〜5。步驟1 0 1 : 針對現有之專利文獻70進行尋找特徵字（英文稱 Mining Term，現已發展出許多著名之技術）。現有之專利文獻70都已經有分類，譬如最有名也被普遍採用的是 IPC ( International Patent Classification，國際專利分類）分類。在進行尋找特徵字時之方式有許多種，譬如現有之方式如透過自然語言處理技術的文法剖析程式，剖析出文件中的名詞片語，再運用一些方法與準則，過濾掉不適合的詞彙。另外亦可將專利文獻7 0出現在詞庫中的片語擷取出來。其他著名之方法如透過對文件的分析，累積足夠的統計參數後，再將統計參數符合某些條件的片語擷取出來。最簡單的統計參數是計數詞彙發生的頻率，即詞頻，將詞頻落在某一範圍的詞彙取出。步驟1 0 2 : 1221992 詞彙關連性運算。譬如在許多專利文獻中，只要出現『3d』這個詞彙，則出現『立體』之機率很高，則代表『3〇』與『立體』為同義子或接近同義。步驟103 : 詞彙集中度運算。譬如在某一 IPC分類中，特徵字出現之頻率的多募。步驟104 : 建立關鍵字類別總表20。關鍵字類別總表2〇可有兩個欄位’技術類別21欄位以及特徵字25欄位。技術類別2 1譬如採用IPC分類（實施例為一示意圖，可比照 IPC分類有五階顯分類），每一技術類別2丨並對應特徵子2 5棚位’特徵字2 5搁位紀錄複數之特徵字2 5。技術類別2 1為『A』的有特徵字『κ e y - A 1』，『Key，A2』，『Key-A3』，『Key_A4』，『Key-A5』，『Key-A6』，『Key_A7』等等。每一特徵字在本實施例之格式可採用：特徵字（比重權值，關連性詞彙_丨，關連性詞囊 -2 ’關連性詞彙，.......關連性詞彙-N ) 譬如特徵字『Key-Al』之比重權值為『80』，且有兩個關連性詞彙，『Key-Al 1』，『Key-A12』。比重權值越大代表此特徵字與對應之技術類別關連性越大。比重權值之計算係由步驟1 〇3所計算分析出來的。而關連性詞彙則是由步驟1 02所計算分析出來的。 10 1221992 請參見圖3關於說明更新關鍵字類別動分類之流程圖，此流程可以讓〜新專利分類，亦可更新關鍵字類別總表2〇。總表2 0及自文獻71自動步驟3 0 1 : 從新專利文獻7 1中尋找特徵字，此 101 〇步驟如步驟步騍3 0 2 : 將新專利文獻7 1的特徵字與P關鍵 2 〇』比較，分析。類別總表/ Weiyu computer system is a fairly well-known device, anyone with a general knowledge of μ · —swears that everyone knows this invention, and the present invention is not changing the computer system, so it will not be repeated here. Function of computer system. The computer system of the present invention includes a patent document database and a search engine. The computer system can use keywords entered by the user to find relevant patent documents. However, the present invention is special in that a keyword category summary table is required in the computer system so that the method of the present invention can work. The following first introduces the generation method of the keyword category summary table, as well as its thinking and meaning. Please refer to FIG. 1 for a flowchart illustrating the generation method of the keyword category master table 20, and please refer to FIGS. 2 to 5 together. Step 101: Find a feature word for the existing patent document 70 (Englishly called Mining Term, and many well-known technologies have been developed). The existing patent documents 70 have been classified. For example, the most famous and commonly used is the IPC (International Patent Classification) classification. There are many ways to find characteristic words. For example, the existing methods such as grammar parsing programs through natural language processing technology, analyze the noun phrases in the document, and then use some methods and guidelines to filter out inappropriate words. In addition, the phrases appearing in the thesaurus of Patent Document 70 can be extracted. Other well-known methods, such as analyzing the files, accumulate enough statistical parameters, and then extract the phrases whose statistical parameters meet certain conditions. The simplest statistical parameter is to count the frequency of vocabulary, that is, the word frequency, and take out the words that fall into a certain range. Step 102: 1221992 lexical relevance calculation. For example, in many patent documents, as long as the word "3d" appears, there is a high probability of "stereo", which means that "3〇" and "stereo" are synonymous or nearly synonymous. Step 103: vocabulary concentration calculation. For example, in a certain IPC classification, the frequency of the appearance of feature words is increased. Step 104: Establish a keyword category master table 20. The keyword category master table 20 may have two fields, a technical category 21 field and a characteristic word 25 field. Technology category 2 1 For example, IPC classification is used (the embodiment is a schematic diagram, which can be compared with the IPC classification with a five-level display classification). Each technology category 2 丨 corresponds to the characteristic sub 2 5 booth 'character word 2 5 Character word 2 5. Technical category 2 1 is the characteristic word "κ ey-A 1" of "A", "Key, A2", "Key-A3", "Key_A4", "Key-A5", "Key-A6", "Key_A7" "and many more. The format of each feature word in this embodiment may be: Feature words (weight weight, related vocabulary _ 丨, related vocabulary-2 'connected vocabulary, ... related vocabulary -N) For example, the weight of the feature word "Key-Al" is "80" and there are two related words, "Key-Al 1" and "Key-A12". The larger the weighting value, the greater the correlation between this feature word and the corresponding technology category. The calculation of the specific weight is based on the calculation and analysis in step 103. The related vocabulary is calculated and analyzed by step 102. 10 1221992 Please refer to Figure 3 for a description of the flow chart of updating the keyword category. This process can be used to classify new patents and update the keyword category summary table 20. General Table 2 0 and Automatic Step 3 0 1 from Document 71: Find the feature word from the new patent document 71. This 101 step is the same as step 骒 3 0 2: The feature word of the new patent document 71 and the P key 2 0. "comparative analysis. Category Summary

步驟3 0 3 : 得出向量表30，如圖4。向量表3〇有兩個為技術類別3 1欄位（即如同關鲮字類別技術類別2 1欄位）以及比重權值3 2攔位。 3 0在技術類別3 1欄位的 A〜D』類所對應 3 2棚位分別為『5 1 2』’『2 0 0 8』，『1 3 因此新專利文獻7 1最有可能是『B』類。向量表3 〇即是在計算比重權值3 2攔位種常見之方式介紹如下，請一併參見：假設新專利文獻7 1的特徵字7 5 a為桌子（5)，辦公（10)，特徵字並對應有一權值的權值為『1 5』。關於特徵字之權值可有計算，譬如出現之頻率，該特徵字出現之出現在發明名稱中權值最大，出現在摘要個攔位，一總表2 0中的譬如向量表之比重權值Step 3 0 3: The vector table 30 is obtained, as shown in FIG. 4. The vector table 30 has two fields of technology category 31 (that is, the same as the key word category technology category 21 field) and a weight of 32. 3 0 in the technical category 3 column A ~ D "corresponding to the 3 2 booths are" 5 1 2 "'" 2 0 0 8 "," 1 3 Therefore the new patent document 7 1 is most likely " Class B '. The vector table 3 〇 is used to calculate the weight of the weight 3 2 common ways are introduced as follows, please also refer to the following: Assume that the characteristic word 7 5 a of the new patent document 7 1 is the table (5), office (10), The feature word corresponds to a weight of "1 5". The weight of the feature word can be calculated, such as the frequency of occurrence, the feature word appears in the invention name with the largest weight value, and appears in the abstract block, a weight in the general table 20, such as the vector table.

之數值，一子（15)，輪，譬如桌子許多方式來地方（譬如或申請專利 11 1221992 範圍中權值次之，其他說明書部分權值最小），此為已知之技術，因此在此僅為舉例。另外假設關鍵字類別總表20a僅有A，B，C三類，而特徵字分別為『椅子（20)，輪子（14)』，『桌子（30)，輪子（1)』，以及『辦公（34)，椅子（10)』。向量表77a的產生係利用簡單之乘法與加法產生。譬如要計算向量表77a在『B』類的比重權值如下： (3 0x 1 5)+ ( 1 x5 )= 455 註：特徵字7 5 a有『桌子』與『輪子』符合關鍵字類別總表20a在『B』類的特徵字。由於『C』類的比重權值為『340』，很接近『B』類的比重權值『45 5』，因此該新專利文獻有可能是『B』類或『C』類。步驟304 : 更新『關鍵字類別總表20』。由於有新專利文獻7 1加入，因此『關鍵字類別總表2 0』可被更新，但一般之作法可等待一定數量之新專利文獻7 1加入後，譬如等每五千筆專利文獻7 1加入後，再更新『關鍵字類別總表20』。以下請參考本發明如何利用『關鍵字類別總表20』以較準確之方式檢索專利的第一實施例。 12 1221992 步驟6 Ο 1 : 接收使用者輸入之搜尋關鍵字。此步驟如同一般之搜尋。步驟602 : 與『關鍵字類別總表20』比較。此步驟如同步驟 301° 步驟603 :The value, one (15), round, such as a table, comes in many ways (for example, the weight is the second in the range of patent application 11 1221992, and the weight in other specifications is the smallest). This is a known technology, so it is only here. For example. In addition, it is assumed that the keyword category general list 20a has only three types: A, B, and C, and the characteristic words are "chair (20), wheel (14)", "table (30), wheel (1)", and "office (34), the chair (10) ”. The vector table 77a is generated using simple multiplication and addition. For example, to calculate the weight of the vector table 77a in the "B" category as follows: (3 0x 1 5) + (1 x5) = 455 Note: The feature word 7 5 a has "table" and "wheel" that match the keyword category total Table 20a is the feature word in category "B". Since the weighting weight of the "C" category is "340", which is very close to the weighting weight of the "B" category "45 5", the new patent document may be a "B" category or a "C" category. Step 304: Update the "Keyword Category General Table 20." As new patent documents 71 are added, the "keyword category master table 20" can be updated, but the general practice can wait for a certain number of new patent documents 7 1 to be added, such as waiting for every five thousand patent documents 7 1 After joining, update "Keyword Category General Table 20". In the following, please refer to the first embodiment of how to use the "Keyword Category Master Table 20" to retrieve patents in a more accurate manner. 12 1221992 Step 6 Ο 1: Receive search keywords entered by the user. This step is like a normal search. Step 602: Compare with "Keyword Category General Table 20." This step is the same as step 301 ° step 603:

得出向量表。此步驟如同步驟302。譬如使用者輸入之搜尋關鍵字為『桌子』與『輪子』，若使用者沒有指定『桌子』與『輪子』出現之處，則可假設『桌子』與『輪子』的權值都為『1』。假設『關鍵字類別總表20』如圖5之關鍵字類別總表20a，則向量表可得出如圖7所示之向量表77b。步驟604 : 分析有可能之類別，其中至少找出一可能之類別 (在申請專利範圍定義為候選技術類別）。馨以步驟603所舉之例子則『B』類（候選技術類別）是最有可能。步驟605 : ~ 依照有可能之類別之專利文獻進行檢索。 - 假設所有的專利文獻總共有一百萬筆，習知技術當輸入關鍵字後，若無輸入尋找『類別』之限制，則 13 1221992 會針對一百萬筆的專利文獻進行全文檢索，因此可能找出非常多無關之資料。由於步驟605等於是幫使用者指定尋找之『類別』，亦即有『類別』之限制，因此可大大縮減無關之資料，譬如降低到尋找一萬筆的專利文獻。Draw the vector table. This step is the same as step 302. For example, if the search keywords entered by the user are "table" and "wheel", if the user does not specify where the "table" and "wheel" appear, it can be assumed that the weights of "table" and "wheel" are both "1" ". Assuming that the "keyword category master table 20" is as shown in the keyword category master table 20a of Fig. 5, the vector table can be obtained as the vector table 77b shown in Fig. 7. Step 604: Analyze the possible categories, and find at least one possible category (defined as a candidate technology category in the scope of patent application). Xin The example given in step 603 shows that the "B" category (candidate technology category) is the most likely. Step 605: ~ Search according to patent documents of possible categories. -Assuming that all patent documents have a total of one million, after entering keywords in the conventional technology, if there is no restriction to search for "category", 13 1221992 will perform a full-text search for one million patent documents, so it is possible Find a lot of irrelevant information. Since step 605 is equivalent to helping the user to specify the "category" to search for, that is to say, there is a restriction on "category", so the irrelevant data can be greatly reduced, for example, it can be reduced to find 10,000 patent documents.

需注意的是步驟605是依照有可能之類別之專利文獻進行檢索，所以可能不只一類，需看向量表得出之結果，譬如若最有可能之類別之權重為『1 00』，則假設只要在最有可能之類別之權重的40%以内都可算為其他有可能之類別，所以任何類別之權重為『40』以上都會被歸為有可能之類別。步驟606 : 顯示結果。此步驟如同一般之搜尋後之顯示結果。以下請參考本發明如何利用『關鍵字類別總表20』以較準確之方式檢索專利之第2實施例，請見圖8。It should be noted that step 605 searches according to patent documents of possible categories, so there may be more than one category, and the results obtained by the vector table are needed. For example, if the weight of the most likely category is "1 00", it is assumed that Within 40% of the weight of the most likely category can be counted as other possible categories, so any category with a weight of "40" or more will be classified as a possible category. Step 606: Display the result. This step is like displaying results after a normal search. Please refer to the second embodiment of the present invention for how to use the "Keyword Category Master Table 20" to retrieve patents in a more accurate manner, see FIG. 8.

在第二實施例中，步驟801〜804與第一實施例之步驟601〜604相同，因此在此不再贅述。步驟8 0 5 : 進行檢索，此步驟如同一般之檢索方式。在第一實施例中，係針對有可能之類別之專利文獻範圍進行檢索，但在第二實施例中，仍如同一般之檢索方式進行檢索。步驟806 : 14 1221992 顯不結果。由於步職？ g Λ 田％乂驟8〇5為一般之檢索方式，所以此步驟可顯示一般之檢奢古切家方式找出之專利文獻，但此步驟由於經過步驟802〜804，因此在顯示結果時可盘 -般傳統方式不同，譬如在顯示找出之專利文獻有註明哪些是屬於候選技術類別之專利文獻（如圖9之檢索結果9〇，使得使用者可以更加注意；或是將越有可能之技術類別之專利文獻排序在前（如圖1〇之檢索結果 101 )等等。以及此步驟之重點在於顯示結果之方^係有參考步驟804分析出有可能之類別。In the second embodiment, steps 801 to 804 are the same as steps 601 to 604 of the first embodiment, so they are not repeated here. Step 805: Perform a search. This step is the same as the normal search method. In the first embodiment, a search is performed for a range of possible patent documents, but in the second embodiment, the search is performed in the same manner as a general search. Step 806: 14 1221992 show no results. Thanks for stepping? g Λ Field% Step 805 is a general search method, so this step can display the patent documents found in the general way of checking luxury and luxury home cutting. However, since this step goes through steps 802 ~ 804, it can be used when displaying the results. Disc-like traditional methods are different. For example, the patent documents found in the display indicate which patent documents belong to the candidate technology category (such as the search result 9 in Figure 9), so that the user can pay more attention; The patent documents of the technology category are ranked first (as shown in the search result 101 of FIG. 10), and so on. And the emphasis of this step is to show the results. Refer to step 804 to analyze possible categories.

本發明 ’而非上述實施例僅係為了方便說明而舉例而已，所主張之權利範圍自應以申請專利範圍所述為準僅限於上述實施例。The present invention ′ is not an example of the foregoing embodiments, but is merely an example for convenience of explanation. The scope of the claimed rights shall be based on the scope of the patent application, and is limited to the above embodiments.

15 1221992 【五、圖式fE 圖1係說明關圖2係說明關圖3係說明更圖。圖4係向量表圖5係說明向圖6係本發明圖7係向量表圖8係本發明圖9係本發明圖10係本發马【圖號說明】關鍵字類別特徵字2 5 專利文獻70 特徵字7 5 a 檢索結果9 1 總表20，20a 技術類別21 向量表30 新專利文獻7 1 向量表77a，77b ，101 ί單說明】鍵字類別總表2 0產生方式之流程圖。鍵字類別總表2 0之實施例。新關鍵字類別總表2 0及自動分類之流程、之實施例。量表產生之實施例。之流程圖第一實施例。之另一實施例。之流程圖第二實施例。籲之流程圖第一實施例。 3之流程圖第二實施例。15 1221992 [fifth, the figure fE Figure 1 shows the description of the figure Figure 2 shows the description of the figure Figure 3 shows the description of the figure. Fig. 4 is a vector table. Fig. 5 is an explanation. Fig. 6 is the present invention. Fig. 7 is the vector table. Fig. 8 is the present invention. Fig. 9 is the present invention. Fig. 10 is the present invention. 70 Feature word 7 5 a Search result 9 1 General table 20, 20a Technical category 21 Vector table 30 New patent document 7 1 Vector table 77a, 77b, 101 Description of the key word type table 2 0 The flow chart of how to generate the general table 2 0. An embodiment of the Key Category General Table 20. The new keyword category summary table 20 and the automatic classification process Example of scale generation. Flow chart of the first embodiment. Another embodiment. Flow chart of the second embodiment. The first embodiment of the flowchart. 3 of the flowchart of the second embodiment.

1616

Claims

1221992 拾、申請專利範圍 _ 1. 一種專利文獻資料檢索之方法，係用於使用者輸入至少一關鍵字於一電腦系統，該電腦系統包括有專 ^ 利文獻資料庫，使得電腦系統尋找出與該關鍵字相關之專利文獻，該方法主要包括下列步驟：步驟A :將該關鍵字與一關鍵字類別總表進行比較，其中關鍵字類別總表係儲存於電腦系統中，關鍵字類別總表紀錄複數之技術類別，以及每一技術類別對應之複數特徵字； β 步驟Β :依照步驟Α由複數之技術類別中分析出該關鍵字至少一有可能屬於之技術類別，在此定義為候選技術類別；步驟C :根據該關鍵字由屬於候選技術類別之專利文獻中進行檢索；以及步驟D :顯示找出之專利文獻。 2 · 如申請專利範圍第1項所述之專利文獻資料檢索之方法，其中關鍵字類別總表之特徵字並對應一比1221992 Scope of patent application and application_ 1. A method for retrieving patent document data, which is used for a user to input at least one keyword into a computer system. The computer system includes a patent literature database, which makes the computer system find out For the keyword-related patent document, the method mainly includes the following steps: Step A: Compare the keyword with a keyword category summary table, where the keyword category summary table is stored in a computer system and the keyword category summary table Record the plural technology categories and the plural feature words corresponding to each technology category; β Step B: According to step A, analyze at least one technology category that the keyword may belong to from the plural technology categories, which is defined here as a candidate technology Category; step C: searching from the patent documents belonging to the candidate technology category according to the keyword; and step D: displaying the found patent documents. 2 · The method of retrieving patent documents as described in item 1 of the patent application scope, in which the feature words of the keyword category master list correspond to one

3. 如申請專利範圍第1項所述之專利文獻資料檢索之方法，在步驟B中找出候選技術類別係利用符合關鍵字之特徵字所對應之比重權值來尋找出。 4. 如申請專利範圍第1項所述之專利文獻資料檢索之方法，其中關鍵字類別總表並記錄與特徵字同義或接近同義之關連性詞彙。 17 1221992 5 · 一種專利文獻資料檢索之方法，係用於使用者 * 輸入至少一關鍵字於一電腦系統，該電腦系統包括有專利文獻負料庫’使得電腦系統尋找出與該關鍵字相關之專利文獻’該方法主要包括下列步驟：步驟A :將該關鍵字與一關鍵字類別總表進行比 - :’其中關鍵字類別總表係儲存於電腦系統中，關鍵 - 子類別總表紀錄複數之技術類別，以及每一技術類別對應之複數特徵字； v驟B ·依照步驟a由複數之技術類別中分析出該關鍵字至少一有可能屬於之技術類別，在此定義為候選技術類別； ' v驟C ’根據該關鍵字進行專利文獻的檢索；以及步驟D :顯示找出之專利文獻，其中顯示之方式係有參考步驟B所找到之候選技術類別。 18 1221992 1 Ο.如申請專利範圍第7項所述之專利文獻資料檢索之方法，其中在步驟D中顯示找出之專利文獻時，係將屬於候選技術類別之專利文獻排序在前。3. According to the method of searching for patent document data described in item 1 of the scope of patent application, the candidate technology category found in step B is found by using the weight weight corresponding to the feature word that matches the keyword. 4. The method for retrieving patent documents as described in item 1 of the scope of patent application, wherein the keyword category summary table records related vocabularies that are synonymous with or close to synonymous features. 17 1221992 5 · A method for retrieving patent literature data, which is used by the user * to input at least one keyword into a computer system, the computer system includes a patent literature negative material library so that the computer system finds the keywords related to the keyword Patent document 'The method mainly includes the following steps: Step A: Compare the keyword with a keyword category summary table-': where the keyword category summary table is stored in a computer system, and the key is a sub-category summary table plural Technology category, and the plural feature word corresponding to each technology category; v Step B · According to step a, analyze at least one technology category that the keyword may belong to from the plurality of technology categories, which is defined as a candidate technology category; 'vStepC' searches for a patent document based on the keyword; and step D: displays the found patent document, where the displayed method is the candidate technology category found with reference to step B. 18 1221992 1 0. The method for retrieving patent documents as described in item 7 of the scope of patent application, wherein when the found patent documents are displayed in step D, the patent documents belonging to the candidate technology category are ranked first.

1919