CN1203411C - Method for searching, accessing and managing data index - Google Patents

Method for searching, accessing and managing data index Download PDF

Info

Publication number
CN1203411C
CN1203411C CN 00133553 CN00133553A CN1203411C CN 1203411 C CN1203411 C CN 1203411C CN 00133553 CN00133553 CN 00133553 CN 00133553 A CN00133553 A CN 00133553A CN 1203411 C CN1203411 C CN 1203411C
Authority
CN
China
Prior art keywords
record
data
collection
search
shelves
Prior art date
Legal status (The legal status 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 status listed.)
Expired - Fee Related
Application number
CN 00133553
Other languages
Chinese (zh)
Other versions
CN1353364A (en
Inventor
林圣富
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
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
Application filed by Individual filed Critical Individual
Priority to CN 00133553 priority Critical patent/CN1203411C/en
Publication of CN1353364A publication Critical patent/CN1353364A/en
Application granted granted Critical
Publication of CN1203411C publication Critical patent/CN1203411C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Landscapes

  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The present invention relates to a method for searching, accessing and managing a data index, which comprises the procedures: 1. data are orderly stored in a data device by the record of G<1>, G<2>, G<3>... G<m>; 2. a record group Gi is read from a random file of a data storage device by a binary searching method; 3. if I<i1>=KW, a procedure 6 is carried out, and if I<i1> is not equal to KW, the procedure 2 is carried out; 4. record data I<i1>, I<i2>, I<i3>... in the record group G<i> are compared with a key word KW; 5. if I<ij>=KW, the procedure 6 is carried out, and otherwise, the method is ended; 6. when I<ij>=KW, data contents are read by R<ij>.

Description

Method for searching, accessing and managing data index
The present invention is meant a kind of hybrid source index, access and/or method for searching, refers to a kind of shelves of sequential file/at random mixed data structure, method of carrying out source index, access/search utilized especially.
Traditional profile is in order to cooperate index/method for searching, can be divided into sequential file (Sequential File) and shelves (Random File) two classes at random haply; Each notes record (Record) of sequential file are when being stored in data storage device, be to deposit in by pen in regular turn, when therefore taking out in the future, also will take out in regular turn, for example to obtain the 20th document, must take out earlier the 1st, 2 in regular turn, 3......19,20 documents, therefore, work efficiency is poor, but because of only need add one separates symbol (for example ASCII character is 13 character) between each document, get final product a document and connect document ground storage, therefore, its storage area less expensive; And shelves are that each notes record all is to be stored in the data storage device with regular length at random, therefore the storage area is uneconomical, but during the access data, as long as indicate which document, get final product direct access, for example want access the 8th document, then its logical bit address is at 7N 10～8N, and wherein N is the regular length of every document.For the huge data bank of data quantity, adopt shelves at random, to the storage area of data bank significantly be risen, be 2 to 90 characters for example to record length, the archives of about 16 characters of average length, Dang archives size approximately is 5.3 times of----90/ (16+1) of depositing the sequential file of identical data at random.For these reasons, present data structure is the data logging storage method of taking to change length (Variablelength) haply, utilizes concordance list (lndex table) to obtain the address of certain document in data bank again.Even so, file mode for concordance list, still run into this selection sequential file storing mode still at random shelves storing mode, present modal mode is to utilize the indexing means of Type B tree (B-tree) to solve, for example referring to the U.S.'s 4945475,5010478,5058002,5204958,5218696,5261088,5265244,5283894,531779,5418947,5430869,5446887,5475837,5522068,5560007,5568638 patents.For the data bank that concordance list changes at any time, the Type B tree truly has its advantage, but the data bank that concordance list is seldom changed, except ordered set (Sequential set), index (lndex set) also will be arranged, referring to Donald E.Knuth., The Art of ComPuterProgramming.Vol III:Sorting and Searching.Reading, Mass:Addison-Welsley (1973)., therefore, its archives size still has more indexed set data content partly than keeping off in proper order.In addition, unless its this Type B tree is complete B-type tree (Complete B-tree), otherwise it searches number of times greater than log 2N (N for record number), that is its searching efficiency is lower than the method for searching of shelves at random, and (the search number of times is log 2N).Therefore, B-type tree can be considered data structure greater than sequential file, and searching efficiency is slightly poorer than the index/method for searching of the binary search method (Binary search) of shelves at random.
With data search, no matter be the index search method of B-type tree, or at random shelves the binary search method, every search once must be carried out the primary access action to data storage device, and the mechanical action of " access ", it often is action the most consuming time, therefore, how reducing the access action in the search process, is the important key that searching efficiency is improved.
For these reasons, the inventor proposes a kind of in searching procedure, can reduce the method for searching of data access number of times, and therefore, this method has the advantage that the raising data is deposited efficient.Moreover, because this method cooperates special hybrid data structure, therefore, with the advantage that reduces the data storage space.
One of the present invention purpose is to provide a kind of data access method.
Another object of the present invention is to provide a kind of data searching method.
Another purpose again of the present invention is to provide a kind of data access index search method.
Another purpose again of the present invention is to provide a kind of hybrid source index, access and/or method for searching.
Source index management method of the present invention, it comprises:
(1) field that will treat index sorts, and makes it be I in regular turn 1, I 2, I 3... I ICalculate its string length respectively, and in regular turn with code L 1, L 2, L 3... L 1Expression;
(2) with the field (I) after step (1) ordering, cooperate its length (L) and location mark (P), form 1 record (R), wherein R in regular turn 1=(L 1I 1P 1), R 2=(L 2I 2P 2), R 3=(L 3I 3P 3) ... R 1=(L 1I 1P 1);
(3) the record R that step (2) is formed 1, R 2, R 3... R 1Form m collection of record G in regular turn and make the total word string length of each collection of record G equal N or less than N, wherein N is the integer greater than 200, and G 1=R 1R 2R 3... R K1, G 2=R K1+1R K1+2... R K2...
Gm＝P km-1+1 R km-1+2......R km；
(4) with an above-mentioned m collection of record, be stored in regular turn in the data storage device, formation one is shelves at random.
Above-mentioned field (Field) is a data (Ij) for the treatment of index, for example the key of encyclopedia retrieval (Searching) system is spoken, the key language of science and technology magazine (Jorunal) searching system, the key language of library's document retrieval system, the key language of Patent data searching system.So-called field length (Length) is meant the word string length of this field, and called site mark (Pointer) is meant the address of this field in data bank, contains its address, shelves name and/or the record position in archives, even record religion.With 1975 to 1996/11 month Patent data storehouses of the U.S. is example, and its keyword string length the shortest is 1 character, growing tall most reaches 480 characters.Referring to No. 5447911 patent of the U.S., this key language is the Amino acid sequence (Aminoacid seqLlece) of peptide (Peptide).Even get rid of protein (Protein), the Amino acid sequence of peptide etc., the longest key language, still contain 104 characters, referring to No. 5140022 patent of the U.S. (3-substituted-aminomethyl-3-substituted-oxy-17a-lower-al kyl-17a-aza-D-homo-5-alpha-androstane), contain " replacement " key language (substituted) even disregard, so that the key language of Special Significance to be arranged, the longest word string, still up to 87 characters, for example referring to No. 5198552 patent of the U.S., this key language 6--bromo-5-hydroxy-4-dimethylaminomethyl-1-methyl-2-phen ylthiomethylindole-3-carboxylate (6-bromo-5-hydroxyl-4-dimethylamino methyl-1-methyl-2-phenyl coloured glaze is for methyl indol-3-carboxylic acid) is meant a specific compound (Compound) or derivatives thereof (derivatives) and its esters (salts) and/or its hydrate (hydrates).United States Patent (USP) with whole year nineteen ninety-five is an example, contain the key language that surpasses 70 characters in its summary (Abstract) and/or the claim (Claim), totally 28, the vinylidene-fluoride-hexafluoropropylene-tetrafluoroethyl ene-bromotetrafluorobutene of No. 5422404 United States Patent (USP) (fluorine time ethene-hexafluoropropylene-tetrafluoroethene-bromine tetrafluoro butylene) for example, the average string length of its crucial language then is 11.7 characters, if add location mark (4 yards) and list separator (totally 4 yards), be saved as sequential file, its archives meter contains 3538427 characters, if be saved as record length is 90 (long 87 characters of crucial language, add the location and mark 3 characters) at random shelves, then its archives character number is 16555680 (90 * 183952), but if be saved as hybrid shelves at random of the present invention, then archives character number is 3061000 (1000 characters * 3061), its archives are less than sequential file, because with respect to sequential file, hybrid shelves at random of the present invention, each record can be saved 4 and separate symbol, deduct the character length code that every notes record must add a character, and 3061 collection of record less than the character number, reduce 477427 characters (average every notes record reduces by 2.6 characters) altogether, referring to following embodiment and comparative example.In view of the above, hybrid-type shelves at random of the present invention, its data storage space is slightly smaller than orthodox retaining in proper order, certainly the data structure storage area of also setting less than the B-type much smaller than orthodox shelves at random.
Above-mentioned hurdle is long Lj, represent the code of the string length of j field Ij, such as: the Ij string length is 10, then defining this code is that the ASCII value is the character of 100 K, wherein K can be any number of 0 to 168 (255-87), K=64 for example, then 10+64=74, so code Lj=" J " (the ASCII value is 74).
Above-mentioned record Rj, be by a data field (Ij), for example the key in the Patent data storehouse is spoken, the word string length code (Lj) of this data, and the location of this data mark (Pj), promptly should key location, the data place mark of language, it may be a certain archive content, this moment, this location mark was to represent this grade name, or certain several record in certain archives, at this moment, this location mark is meant this grade name and rises and write down the location mark so far, or number is marked and write down altogether in the home record location.
Above-mentioned collection of record is made up of plurality of records, and the contained record number of each collection of record may be identical, also may be different, and for example its constituted mode is:
G 1＝R 1R 2R 3......R k1
G 2＝R k1+1R k1+2R k1+3......R k2
G 3＝P k2+1R k2+2R k2+3...............R k3
.............................................
Gm＝R km-1+1R km-1+2R km-1+3......R km
Collection of record wherein, G 2, G 3... G mContain k respectively 1, k 2-k 1, k 3-k 2... n-k N-1Individual record, and k 1, k 2-k 1, k 3-k 2... n-k N-1Value not necessarily equate, fully with G 1, G 2, G 3... G mWhether total word string length equals N or be slightly smaller than N is basis for estimation, certainly G mBe made of remaining record, its total word string length may be much smaller than N.The number of N, except restriction, there is no other restrictions, but be the best with 2 times of word string length of the longest Rj word string length to the computer primary access more than or equal to the longest Rj word string length, for better, is the best with 20 times to 200 times with 10 to 300 times of the longest Rj word string length.When G1 word string length during greater than the word string length of computer primary access, then collection of record of access must secondary or the above access of secondary, and this makes the access time of mechanical action increase, and is unfavorable for the access/index/searching efficiency of data.When the one notes record of computer primary access, can utilize array technique, cooperate each Rj collection of record of its contained record to be reduced into a plurality of R j, for example with G 3Be reduced into array R K2+1+ R K2+2+ R K2+3... R K3, because homing action is homing action in proper order, so K 1, K 2, K 3... K mValue can not be too big so that the mechanical type that the homing action time is less than or equal to is once deposited the time.
Above-mentioned data storage device, it can be arbitrarily known data storage device, Winchester disk drive for example, floppy drive/disc, CD player/CD, read-write formula CD player etc.
By above-mentioned comparison, data structure of the present invention as can be known, its archives size is much smaller than traditional shelves at random, also less than traditional sequential file, certainly still less in the index shelves data shelves of Type B tree.And this data structure, if cooperate following data search access method of the present invention, then it searches access speed also than traditional sequential file Type B tree or shelves are also fast at random.
Data search access method of the present invention, it comprises:
(1) with collection of record G 1, G 2, G 3... G mIn regular turn data is existed in the data storage device, formation one is shelves at random, wherein G 1, G 2, G 3... G mRespectively by K 1, K 2, K 3... K mIndividual record R IjConstitute, and each the record R IjAt least contain data length (L Ij), data (I Ij) and location mark (P Ij), wherein i=1～m, j=1～ki;
(2) with the binary search method, by data storage device this at random shelves in, reading and recording group Gi, and with this collection of record first the record in data (I I1) conduct and key to be searched language KW compare;
(3) if I I1=KW then carries out step (6), if I I1≠ KW then according to step (2), continues to search to I I1=KW carries out step (6) again, or searches to collection of record G iFirst the record in data I I1＜Kw, and collection of record G I+1First the record in data I I+1＞KW carries out step (4) again;
(4) with collection of record G iIn each the record in data I I1, I I2, I I3... and crucial language KW compares;
(5) if I Ij=KW then carries out step (6), if I I1, I I2, I I3... all be not equal to crucial language KW, the expression index in the shelves, does not meet the data of searching crucial language KW at random, and indication does not have and finishes after meeting data;
(6) I IjDuring=KW, then by R IjRead the data content of waiting to search crucial language.
In above-mentioned steps (1)～(6), the subscript of collection of record G is represented the sequence number of this collection of record, for example G 1, G 2... G i... represent the 1st, 2 respectively ... i... collection of record G; k 1, k 2, k 3... k mBe respectively collection of record G 1, G 2, G 3... G mIn contained record (R Ij) number, wherein the subscript of the subscript of K and collection of record becomes man-to-man correspondence, i.e. k 1-G 1, k 2-G 2...; R IjFirst subscript represent I collection of record (G i), and on behalf of this, the second subscript j be recorded as collection of record G iJ record; I Ij, L Ij, P IjSubscript define same R Ij, represent i collection of record G respectively iJ data (I Ij), data length (L Ij) and location mark (P Ij).
Above-mentioned data length (L Ij), data (I Ij), location mark (P Ij), field (R j) and collection of record (G i) relation, roughly the same the 6th page, the 14th to walk to the 8th page of the 2nd row described.
The binary search method of above-mentioned steps (2), (3) is known binary search method or its improvement.
In the above-mentioned steps (4), I I1, I I2, I I3... and the search relative method of KW, can adopt sequential search or binary search method.In above-mentioned data structure, with G iResolve into R I1, R I2, R I3..., and then single from going out I I1, I I2, I I3... program itself be to adopt in proper order to decompose, when therefore decomposing in proper order, carry out sequential search simultaneously relatively for preferable.Above-mentioned decomposition method in proper order, for example: by G iL I1, can know I by inference I1(second character to the L I1+ 1 character) and L I2(L I1+ n+2 character, wherein n is P IjUnified character number), by L I+2Can know I by inference I2And L I3, in like manner, can be pushed into I I3, L I4, I I4, L I5... etc.Data structure of the present invention also can cooperate and is modified to: Gi=K iQ I1Q I2Q I3... Q IkR I1R I2... R Ik
Wherein k represents total k record among the Gi, Q I1, Q I2, Q I3... Q IkRepresent R respectively I1, R I2, R I3... R IkThe start character of record, and record Rij is with above-mentioned, i.e. R Ij=L IjI IjP IjIf with above-mentioned data structure, then in the step (4), I IjWith the comparative approach of KW, can adopt the binary search method, to speed its search speed.
For further specifying the advantage of the inventive method, now (invent 5377359-5479657 number with U.S.'s nineteen ninety-five patent, appearance design D353932-D365670 number, H1389-H1511 number, PP9026-PP9412 number, and RE34814～RE35135 number) be example, utilize traditional sequential file, shelves and data structure of the present invention at random, relatively (CPU of computer is 486DX-100 for its data storage space and search speed, program is write as with Quick-Basic, and shelves are carried out through being compiled in the end) relatively.
Below in conjunction with embodiment and accompanying drawing technical scheme of the present invention is further described.
The operational flowchart that Fig. 1 and Fig. 2 are respectively traditional random access and follow the example of in proper order.
Fig. 3 is the operational flowchart of access method one preferred embodiment of the present invention.
95.RND among Fig. 1 to Fig. 3,95.SQN and 95.INV are file name, see the explanation of comparative example 1,2 and embodiment 1 for details.
Comparative example 1:
183952 crucial speaking with U.S.'s nineteen ninety-five patent with I jP jForm be saved as at random shelves, wherein I jAccount for 87 bit groups, P jAccount for 3 bit groups, the archives size is 16555680 bit groups, (90X183952) shelves 95.RND by name.
Search (searching 1000 Kw) according to following flow diagram (Fig. 1) at every turn, test five times, its result is respectively T=33.28906 second, and 33.67188 seconds, 33.66797 seconds, 32.35156 seconds and 32.73828 seconds.
Comparative example 2:
Roughly the same comparative example 1, with 183352 crucial languages of U.S.'s nineteen ninety-five patent, with R j=I j, P jForm is saved as sequential file, wherein I jBe word string (string), and P jBe integer, this archives size is 358427 bit groups, shelves 95.SQN by name.
According to the test of seeing off of flow process method shown in Figure 2, the search times that record 100 crucial languages are 753.25 seconds.
Embodiment 1.
With 183952 crucial languages of U.S.'s nineteen ninety-five patent, with R Ij=L IjI IjP IjMode, gather into the collection of record G that a length equals or be bordering on 1000 bit groups in regular turn i, in the mode of each collection of record 1000 bit group it being saved as shelves at random, its grade size is 3061000 bit groups, shelves 95.INV by name.
According to method shown in the following flow diagram (Fig. 3), to the 95.INV test 10 times of seeing off, the search times that record 100 crucial languages are respectively T=2.410156 second, 2.640625 seconds, 2.46875 second, 2.519531 second, 2.476563 seconds, 2.582031 seconds, 2.578125 second, 2.53125 second, 2.359375 seconds and 2.371094 seconds.
Conclude comparative example 1.2 and embodiment one, the result is as follows:
Archives size (bit group) Average every crucial language search time
Sequential file 3538427 About 7.53 seconds
At random shelves 16555680 0.032～0.034 second
The present invention 3061000 0.024～0.027 second
As seen from the above table, the data storage space minimum of the inventive method, and search time is the shortest, speed is fast.

Claims (7)

1. source index management method, it comprises:
(1) field that will treat index sorts, and makes it be I in regular turn 1, l 2, l 3... I 1Calculate its string length respectively, and in regular turn with code L 1, L 2, L 3... L 1Expression;
(2) with the field I after step (1) ordering, cooperate its length L and location mark P, form 1 record R, wherein R in regular turn 1=L 1I 1P 1, R 2=L 2I 2P 2, R 3=L 3I 3P 3... ... R 1=L 1I 1P 1
(3) the record R that step (2) is formed 1, R 2, R 3... R 1Form m collection of record G in regular turn and make the total word string length of each collection of record G equal N or less than N, wherein N is the integer greater than 200, and G 1=R 1R 2R 3... Rk 1, G 2=Rk 1+ 1Rk 1+ 2Rk 2... Gm=Rk M-1+ 1Rk M-1+ 2......Rk m
(4) with an above-mentioned m collection of record, be stored in regular turn in the data storage device, formation one is shelves at random.
2. source index management method as claimed in claim 1, wherein this N value is that 2 times of the longest record Rj word string length are to computer primary access word string length.
3. source index management method as claimed in claim 2, wherein this N value is 10 to 300 times of the longest record Rj word string length.
4. source index management method as claimed in claim 3, wherein 20 to 200 times of the longest record of this N value Rj word string length.
5. as claim 1,2,3 or 4 described source index management methods, wherein respectively the forefront of this collection of record adds K iQ I1Q I2Q I3... Q IkR I1R I2R I3... R IkWherein k represents total k record among the Gi, Q I1Q I2Q I3... Q IkRepresent R respectively I1, R I2, R I3... R IkThe start character of record.
6. data search access method, it comprises
(1) with record G 1, G 2, G 3... G m, in regular turn data being existed in the data device, formation one is shelves at random, wherein G 1, G 2, G 3... G mRespectively by K 1, K 2, K 3... K mIndividual record Rij constitutes, and each record R IjAt least contain the data length L Ij, data I IjWith location mark P Ij, wherein i=1～m, j=1～ki;
(2) with the binary search method, by data storage device this at random shelves in, reading and recording group Gi, and with this collection of record first the record in data I I1As comparing with key language KW to be searched;
(3) if I I1=KW then carries out step (6), if I I1≠ KW then according to step (2), continues to search to I I1=KW carries out step (6) again, or search to collection of record Gi first the record in data I I1＜Kw, and collection of record G I+1First the record in data I I+1＞KW carries out step (4) again;
(4) with collection of record G iIn each the record in data I I1, I I2, I I3... and crucial language KW compares;
(5) if I Ij=KW then carries out step (6), if I I1, I I2, I I3... all be not equal to crucial language KW, the expression index in the shelves, does not meet the crucial more data of KW of searching at random, and indication does not have and finishes after meeting data;
(6) I IjDuring=KW, then by R IjRead the data content of waiting to search crucial language.
7. data search access method as claimed in claim 6, wherein respectively the forefront of this collection of record adds MQ 1Q 2Q 3... Q mR 1R 2R 3... R mWherein k represents G iIn total k record, Q I1, Q I2, Q I3... Q IkRepresent R respectively I1, R I2, R I3... R IkThe start character of record.
CN 00133553 2000-11-10 2000-11-10 Method for searching, accessing and managing data index Expired - Fee Related CN1203411C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 00133553 CN1203411C (en) 2000-11-10 2000-11-10 Method for searching, accessing and managing data index

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 00133553 CN1203411C (en) 2000-11-10 2000-11-10 Method for searching, accessing and managing data index

Publications (2)

Publication Number Publication Date
CN1353364A CN1353364A (en) 2002-06-12
CN1203411C true CN1203411C (en) 2005-05-25

Family

ID=4595797

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 00133553 Expired - Fee Related CN1203411C (en) 2000-11-10 2000-11-10 Method for searching, accessing and managing data index

Country Status (1)

Country Link
CN (1) CN1203411C (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI480744B (en) * 2008-06-13 2015-04-11 Microsoft Corp Search index format optimizations

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100382081C (en) * 2004-11-26 2008-04-16 中兴通讯股份有限公司 History data wrapped connection

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI480744B (en) * 2008-06-13 2015-04-11 Microsoft Corp Search index format optimizations

Also Published As

Publication number Publication date
CN1353364A (en) 2002-06-12

Similar Documents

Publication Publication Date Title
CN1020014C (en) Method of rapidly opening disk files identified by path names
US8131726B2 (en) Generic architecture for indexing document groups in an inverted text index
WO2013152678A1 (en) Method and device for metadata query
WO2007141809A1 (en) Data mining using an index tree created by recursive projection of data points on random lines
WO2006081325A2 (en) Multiple index based information retrieval system
US7783589B2 (en) Inverted index processing
CN105912687A (en) Mass distributed database memory cell
CN103229164A (en) Data access method and device
CN1776688A (en) Word data searching method
US20070094313A1 (en) Architecture and method for efficient bulk loading of a PATRICIA trie
Baeza-Yates et al. Hierarchies of indices for text searching
CN101751517B (en) Method and system for fast processing genome short sequence mapping
KR20170065374A (en) Method for Hash collision detection that is based on the sorting unit of the bucket
CN1345426A (en) System and method for extracting index key data fields
CN1146819C (en) Computer file title searching system
CN1203411C (en) Method for searching, accessing and managing data index
CN1286043C (en) Method of quickly locating record in data gage in data base
US20240220470A1 (en) Data storage device and storage control method based on log-structured merge tree
WO2011073680A1 (en) Improvements relating to hash tables
Putz Using a relational database for an inverted text index
US20110231404A1 (en) File storage and retrieval method
CN1287316C (en) Method and system for compressing column becoming longer in period of indexing high key code generation
CN113742307A (en) Secondary index storage and query method and system based on value log system
CN1492363A (en) Data storage and searching method of embedded system
EP0567668A1 (en) A computer system for retrieval of information

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C14 Grant of patent or utility model
GR01 Patent grant
C19 Lapse of patent right due to non-payment of the annual fee
CF01 Termination of patent right due to non-payment of annual fee