CN103902699B - Data space retrieval method applied to big data environments and supporting multi-format feature - Google Patents

Data space retrieval method applied to big data environments and supporting multi-format feature Download PDF

Info

Publication number
CN103902699B
CN103902699B CN201410125840.0A CN201410125840A CN103902699B CN 103902699 B CN103902699 B CN 103902699B CN 201410125840 A CN201410125840 A CN 201410125840A CN 103902699 B CN103902699 B CN 103902699B
Authority
CN
China
Prior art keywords
index
query
key word
keyword
item
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
CN201410125840.0A
Other languages
Chinese (zh)
Other versions
CN103902699A (en
Inventor
周连科
王洪滨
王念滨
祝官文
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.)
Harbin Engineering University
Original Assignee
Harbin Engineering University
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 Harbin Engineering University filed Critical Harbin Engineering University
Priority to CN201410125840.0A priority Critical patent/CN103902699B/en
Publication of CN103902699A publication Critical patent/CN103902699A/en
Application granted granted Critical
Publication of CN103902699B publication Critical patent/CN103902699B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/22—Indexing; Data structures therefor; Storage structures
    • G06F16/2228—Indexing structures
    • G06F16/2272—Management thereof

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention relates to a data space retrieval method applied to big data environments and supporting multi-format feature. The method includes: inputting query content by a user; judging the type of query from the user; reading a built multi-stage index by means of prefix scanning; merging linked lists; rewriting query; traversing the multi-stage index; pressing an inverted sequence into a stack; popping two elements from the stack top first; reading the built multi-stage index; popping out stack top elements according to an index list table right-join scheme; outputting all elements which are satisfactory. The method has the advantages that the multi-stage index is built with a B-tree index and a secondary index and the problem that the cost of a main index on path query index joins in the big data environments can be solved.

Description

The data space that multi-format characteristic can be supported under a kind of environment for big data is retrieved Method
Technical field
The present invention relates to the data space search method of multi-format characteristic can be supported under a kind of environment for big data.
Background technology
Data space data have diversified feature, and it includes structuring, such as semi-structured and unstructured data, relation The data such as first ancestral, XML, Word document, Email, video, audio frequency, this feature causes to support polytype in the urgent need to a kind of Inquiry mode, therefore its index technology just seems particularly critical.On the one hand, with search engine and traditional data integrated technology not With data space index technology demand draws multiple types of data, rather than sets up a kind of index for each type；The opposing party Face, is no longer the inquiry for laying particular emphasis on certain categorical data and unlike traditional search engines, XML engines, data base querying, But the inquiry of various different degrees of structures is neatly supported, such as keyword query, predicate inquiry, path query.
As developing rapidly for the Internet, data message are presented explosive growth, every year the new data of at least more than one hundred million TB is produced Raw, in the face of this big data environment, the height of its index efficiency, the performance of direct determination data Spatial data query is fine or not, because This, the index efficiency of data space data is very crucial.At present data space index technology mainly have Hybrid-ATIL indexes, Index of the picture, full-text index+copy, although these index technologies have indexed well multiple types of data, they are difficult to solve Certainly the data space under big data environment indexes the low problem of joint efficiency.For the problem, the present invention utilizes multiple index Thought, have devised the data space efficient index technology under a kind of environment for big data, so as to improve query performance.
The content of the invention
It is an object of the invention to provide a kind of support various inquiry modes and index multi-format data, can reduce a large amount of Attended operation, the efficient data space search method for being used under big data environment that multi-format characteristic can be supported.
The object of the present invention is achieved like this：
The data space search method of multi-format characteristic can be supported under a kind of environment for big data, including：
1) user input query content；
2) user's query type is judged, if keyword query Q={ ki, kiFor key word, then execution step 3)；Such as Fruit inquires about Q=(v, { k for predicatei), v is attribute, kiFor key word, then step 5 is gone to)；If path query Q= k1/..../ki/ ... '/' representational level path, then go to step 7)；
3) set up multiple index is read using prefix scan mode, obtains ki*, ki* represent with key word kiStart rope Draw item, start corresponding chained list result and be designated as respectivelyRepresent index in j-th include kiText corresponding to index entry Shelves list, i.e. posting；If query type is keyword query, step 4 is gone to)；If query type is looked into for path Ask, then go to step 7)；
4) chained list union operation is carried out, i.e.,First to all kiThe corresponding posting of index entry of beginning Carry out and operate, to all key word kiAnd after result carry out friendship operation, while there is the lists of documents of all key words；
5) query rewrite is { ki//v}；
6) traversal step 1) multiple index set up, obtain kiThe corresponding items of //v, are designated asLki//vTable Show k in index entryi//v correspondence posting, L are represented in attribute while there are multiple key word kiAll lists of documents；
7) by k1To knIn backward press-in stack；
8) two elements in stack top are ejected first, be designated as k1And k2；
9) read step 1) multiple index set up, obtain k1B- trees index and k2H indexes, be designated as respectively Key word k1Correspondence resource view numbering be element constitute b-tree indexed andKey word k2Corresponding H indexes；
10) according to the right connection scheme of index chained list, connectionWithAs a result it is designated asThe interim B for generating Tree, is initially empty, i.e., rightIn each major keyFor, ifIn it can be found that, thenRespective items C={ ciIn All elements are inserted into BtempIn；
11) if stack is not sky, step 12 is gone to)；Otherwise, step 14 is gone to) in；
12) stack top element k is ejectedi, read step 1) and the multiple index set up, obtain kiH indexes, be designated asPress According to step 10) method connectionAnd Btemp, as a result it is designated as
13) step 11 is gone to)；
14) B is traveled throughtempOr L, export all elements for meeting condition.
Step 3) building process of multiple index set up comprises the steps：
A, under big data environment to data space build multiple index；The content construction includes arranging rope using extension Structure of the structure of the master index for drawing with the secondary index using B- trees in combination with secondary index；Extension inverted index is responsible for propping up Hold keyword query, the predicate inquiry of big data；B- trees index and secondary index is then responsible for supporting the path query of big data；
The structure of master index is for different components in resource view, using the inverted index of extension structure to be indexed Build：
(1) loading data spatial data；
(2) to each resource view ViEntitled keyword title component, key word row in add keyword, And item (V is added in corresponding chained listi,{Pi k), wherein ViRepresent resource view ViUnique mark, { Pi kRepresent Vk→Vi All VkMark constitute set；That is ViAll father nodes mark；
(3) if the tuple component of resource view is not sky, step 4 is gone to), if content components are not sky, turn To step (5)；
(4) to tuple component τ=(w, t) of resource view, wherein, w intermediate schemes, t is a unit for meeting pattern w Group；W=aj, j=1,2 ..k is a sequence of attributes, wherein ajFor attribute-name；T=vj, j=1,2 ... k are value sequences, wherein vjTo be worth, this step includes two sub-steps (4-1) and (4-2)；
(4-1) a is added in key word rowjAttribute-name, and item (V is added in corresponding chained listi,{Vi), wherein ViTable Show resource view ViUnique mark；
(4-2) ＜ a, k ＞ is a corresponding attribute-value pair of (w, t), then to each ＜ a, k ＞, in key word K//a is added in row, and adds an item (V in corresponding chained listi,{Vi), wherein ViRepresent resource view ViUnique mark Know；
(5) for each key word keyword in content components, keyword is added in key word row, and in phase Answer one item (V of addition in chained listi,{Vi), wherein ViRepresent resource view ViUnique mark；
B, secondary index are mainly to solve master index under big data environment, and path query index connection is of a high price Problem；Secondary index is made up of B- trees index and secondary index, and it is comprised the following steps that：
(1) master index is read；
(2) to each key word keyword1, its corresponding item term is obtainedi=..., ＜ Vi,{Pi k＞ ..., Pi kIt is viParent resource view；
(3) if keyword1 is not expanded keyword, i.e. a//k forms then carry out following two steps：
(3-1) S={ V are assumedi, wherein ViIt is comprising key word for the left-half of all elements of item i, i.e. S All resource views of keyword1, then carry out B- trees index to S；
(3-2) to each element ＜ V in item ii,{Pi kEach P in ＞i kIf not including P in father view vectori k, In being then added to father view vector, and ViIn being added to its corresponding chained list, H indexes are formed.
Multilevel index technology, using the different types of inquiry of right linking rule process.
The beneficial effects of the present invention is：
The method of the present invention, by B- trees index and secondary index multiple index is collectively formed, and can solve the problem that master index big Under data environment, the excessive problem of path query index connection cost.
Description of the drawings
Fig. 1 is the querying method based on the right linking rule of major-minor index；
Fig. 2 is the right connection scheme of index chained list；
Fig. 3 is the secondary index method for combining B- trees and secondary index.
Fig. 4 is the B-tree and H tree examples that key word keyword1 and keyword2 are applied
Specific embodiment
The present invention is described further below in conjunction with the accompanying drawings.
1) user input query content；
2) user's query type is judged, if keyword query Q={ ki, then go to step 3)；Look into if predicate Ask Q=(v, { ki), then go to step 5)；If path query Q=k1/..../ki/ ..., then go to step 7).
3) set up multiple index is read using prefix scan mode, obtains ki* corresponding chained list result difference is started It is designated asIf query type is keyword query, step 4 is gone to)；If query type is path query, turn To step 7)；
4) chained list union operation is carried out, i.e.,
5) query rewrite is { ki//v}；
6) traversal step 1) multiple index set up, obtain kiThe corresponding items of //v, are designated as
7) by k1To knIn backward press-in stack；
8) two elements in stack top are ejected first, be designated as k1And k2；
9) read step 1) multiple index set up, obtain k1B- trees index and k2H indexes, be designated as respectivelyWith
10) according to the right connection scheme of index chained list, connectionWithAs a result it is designated asIt is initially empty, It is i.e. rightIn each major keyFor, ifIn it can be found that, thenRespective items C={ ciIn all elements insertion To BtempIn；
11) if stack is not sky, step 12 is gone to)；Otherwise, step 14 is gone to) in；
12) stack top element k is ejectedi, read step 1) and the multiple index set up, obtain kiH indexes, be designated asPress According to step 10) method connectionAnd Btemp, as a result it is designated as
13) step 11 is gone to)；
14) B is traveled throughtempOr L, export all elements for meeting condition.
Present embodiment is illustrated with reference to Fig. 1 to Fig. 3, for multi-format characteristic can be supported under big data environment Data space search method, methods described comprises the steps (as shown in Figure 1)：
1) user input query content；
2) user's query type is judged, if keyword query Q={ ki, then go to step 3)；Look into if predicate Ask Q=(v, { ki), then go to step 5)；If path query Q=k1/..../ki/ ..., then go to step 7).
3) set up multiple index is read using prefix scan mode, obtains ki* corresponding chained list result difference is started It is designated asIf query type is keyword query, step 4 is gone to)；If query type is path query, Go to step 7)；
4) chained list union operation is carried out, i.e.,
5) query rewrite is { ki//v}；
6) traversal step 1) multiple index set up, obtain kiThe corresponding items of //v, are designated as
7) by k1To knIn backward press-in stack；
8) two elements in stack top are ejected first, be designated as k1And k2；
9) read step 1) multiple index set up, obtain k1B- trees index and k2H indexes, be designated as respectivelyWith
10) according to the right connection scheme (as shown in Figure 2) of index chained list, connectionWithAs a result it is designated asBe initially empty, i.e., it is rightIn each major keyFor, ifIn it can be found that, thenRespective items C={ ciIn all elements be inserted into BtempIn；
11) if stack is not sky, step 12 is gone to)；Otherwise, step 14 is gone to) in；
12) stack top element k is ejectedi, read step 1) and the multiple index set up, obtain kiH indexes, be designated asPress According to step 10) method connectionAnd Btemp, as a result it is designated as
13) step 11 is gone to)；
14) B is traveled throughtempOr L, export all elements for meeting condition.
Step 3) building process of multiple index set up includes following A step and step B：
A, under big data environment to data space build multiple index；The content construction includes arranging rope using extension Structure of the structure of the master index for drawing with the secondary index using B- trees in combination with secondary index；Wherein, inverted index is extended It is responsible for supporting keyword query, the predicate inquiry of big data；B- trees index and secondary index is then responsible for supporting the path of big data Inquiry；
The structure of master index is mainly, for different components in resource view, to be indexed using the inverted index of extension Build, its process is as follows：
(1) loading data spatial data；
(2) to each resource view ViEntitled keyword title component for, key word row in add Keyword, and item (V is added in corresponding chained listi,{Pi k), wherein ViRepresent resource view ViUnique mark, { Pi kTable Show Vk→Vi(Pi kIt is viParent resource view) all VkMark constitute set；That is ViAll father nodes mark；
(3) if the tuple component of resource view is not sky, step 4 is gone to), if content components are not sky, turn To step (5)；
(4) for tuple component τ=(w, t) of resource view, wherein, w intermediate schemes, t is meet pattern w one Tuple；W=aj, j=1,2 ..k is a sequence of attributes, wherein ajFor attribute-name；T=vj, j=1,2 ... k are value sequences, its Middle vjTo be worth, this step includes two sub-steps (4-1) and (4-2)；
(4-1) a is added in key word rowj, and item (V is added in corresponding chained listi,{Vi), wherein ViRepresent resource View ViUnique mark；
(4-2) assume that ＜ a, k ＞ are a corresponding attribute-value pair of (w, t), then to each ＜ a, k ＞, are closing K//a is added in keyword row, and adds an item (V in corresponding chained listi,{Vi), wherein ViRepresent resource view ViOnly One mark；
(5) for each key word keyword in content components, keyword is added in key word row, and in phase Answer one item (V of addition in chained listi,{Vi), wherein ViRepresent resource view ViUnique mark；
B, secondary index are mainly to solve master index under big data environment, and path query index connection is of a high price Problem；Secondary index is made up of B- trees index and secondary index, with reference to Fig. 3, illustrates that it is comprised the following steps that：
(1) master index is read；
(2) to each key word keyword1, its corresponding item term is obtainedi=..., ＜ Vi,{Pi k＞ ...；
(3) if keyword1 is not expanded keyword, i.e. a//k forms then carry out following two steps：
(3-1) S={ V are assumedi, wherein ViIt is comprising key word for the left-half of all elements of item i, i.e. S All resource views of keyword1, then carry out B- trees index to S；
(3-2) to each element ＜ V in item ii,{Pi kEach P in ＞i kIf not including P in father view vectori k, In being then added to father view vector, and ViIn being added to its corresponding chained list, H indexes are formed.

Claims (3)

1. the data space search method of multi-format characteristic can be supported under a kind of environment for big data, it is characterised in that：
1) user input query content；
2) user's query type is judged, if keyword query Q={ ki, kiFor key word, then execution step 3)；If Predicate inquires about Q=(v, { ki), v is attribute, kiFor key word, then step 5 is gone to)；If path query Q=k1/..../ ki/ ... '/' representational level path, then go to step 7)；
3) set up multiple index is read using prefix scan mode, obtains ki*, ki* represent with key word kiStart index , start corresponding chained list result and be designated as respectivelyRepresent index in j-th include kiDocument corresponding to index entry List, i.e. posting；If query type is keyword query, step 4 is gone to)；If query type is path query, Then go to step 7)；
4) chained list union operation is carried out, i.e.,First to all kiThe corresponding posting of index entry of beginning is carried out And operate, to all key word kiAnd after result carry out friendship operation, while there is the lists of documents of all key words；
5) query rewrite is { ki//v}；
6) traversal step 3) multiple index set up, obtain kiThe corresponding items of //v, are designated asLki//vRepresent rope Draw k in itemi//v correspondence posting, L are represented in attribute while there are multiple key word kiAll lists of documents；
7) by k1To knIn backward press-in stack；
8) two elements in stack top are ejected first, be designated as k1And k2；
9) read step 3) multiple index set up, obtain k1B- trees index and k2H indexes, be designated as respectivelyKey word k1Correspondence resource view numbering be element constitute b-tree indexed andKey word k2Corresponding H indexes；
10) according to the right connection scheme of index chained list, connectionWithAs a result it is designated asThe interim B-tree for generating, just Begin as sky, i.e., it is rightIn each major keyFor, ifIn it can be found that, thenRespective items C={ ciIn own Element is inserted into BtempIn；
11) if stack is not sky, step 12 is gone to)；Otherwise, step 14 is gone to) in；
12) stack top element k is ejectedi, read step 3) and the multiple index set up, obtain kiH indexes, be designated asAccording to step Rapid 10) method connectionAnd Btemp, as a result it is designated as
13) step 11 is gone to)；
14) B is traveled throughtempOr step 6) all lists of documents L, output meets all elements of condition.
2. the data space retrieval side of multi-format characteristic can be supported under a kind of environment for big data according to claim 1 Method, it is characterised in that：The step 3) building process of multiple index set up comprises the steps：
A, under big data environment to data space build multiple index；The content construction is included using extension inverted index Structure of the structure of master index with the secondary index using B- trees in combination with secondary index；Extension inverted index is responsible for supporting big The keyword query of data, predicate inquiry；B- trees index and secondary index is then responsible for supporting the path query of big data；
The structure of master index is for different components in resource view, using the inverted index of extension structure to be indexed：
(1) loading data spatial data；
(2) to each resource view ViEntitled keyword title component, key word row in add keyword, and Item (V is added in corresponding chained listi,{Pi k), wherein ViRepresent resource view ViUnique mark, { Pi kRepresent Vk→ViInstitute There is VkMark constitute set；That is ViAll father nodes mark；
(3) if the tuple component of resource view is not sky, step 4 is gone to), if content components are not sky, go to step Suddenly (5)；
(4) to tuple component τ=(w, t) of resource view, wherein, w intermediate schemes, t is a tuple for meeting pattern w；W= aj, j=1,2 ..k is a sequence of attributes, wherein ajFor attribute-name；T=vj, j=1,2 ... k are value sequences, wherein vjFor Value, this step includes two sub-steps (4-1) and (4-2)；
(4-1) a is added in key word rowjAttribute-name, and item (V is added in corresponding chained listi,{Vi), wherein ViRepresent money Source view ViUnique mark；
(4-2) ＜ a, k ＞ is a corresponding attribute-value pair of (w, t), then to each ＜ a, k ＞, in key word row K//a is added, and adds an item (V in corresponding chained listi,{Vi), wherein ViRepresent resource view ViUnique mark；
(5) for each key word keyword in content components, keyword is added in key word row, and in corresponding chain Add an item (V in tablei,{Vi), wherein ViRepresent resource view ViUnique mark；
, mainly to solve master index under big data environment, path query index connection is of a high price to ask for B, secondary index Topic；Secondary index is made up of B- trees index and secondary index, and it is comprised the following steps that：
(1) master index is read；
(2) to each key word keyword1, its corresponding item term is obtainedi=..., ＜ Vi,{Pi k＞ ..., Pi kIt is viParent resource view；
(3) if keyword1 is not expanded keyword, i.e. a//k forms then carry out following two steps：
(3-1) S={ V are assumedi, wherein ViIt is comprising key word keyword1 for the left-half of all elements of item i, i.e. S All resource views, then carry out B- trees index to S；
(3-2) to each element ＜ V in item ii,{Pi kEach P in ＞i kIf not including P in father view vectori k, then add To in father view vector, and ViIn being added to its corresponding chained list, H indexes are formed.
3. the data space retrieval side of multi-format characteristic can be supported under a kind of environment for big data according to claim 1 Method, it is characterised in that described multilevel index technology, using the different types of inquiry of right linking rule process.
CN201410125840.0A 2014-03-31 2014-03-31 Data space retrieval method applied to big data environments and supporting multi-format feature Expired - Fee Related CN103902699B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410125840.0A CN103902699B (en) 2014-03-31 2014-03-31 Data space retrieval method applied to big data environments and supporting multi-format feature

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410125840.0A CN103902699B (en) 2014-03-31 2014-03-31 Data space retrieval method applied to big data environments and supporting multi-format feature

Publications (2)

Publication Number Publication Date
CN103902699A CN103902699A (en) 2014-07-02
CN103902699B true CN103902699B (en) 2017-04-12

Family

ID=50994021

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410125840.0A Expired - Fee Related CN103902699B (en) 2014-03-31 2014-03-31 Data space retrieval method applied to big data environments and supporting multi-format feature

Country Status (1)

Country Link
CN (1) CN103902699B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104133890B (en) * 2014-07-30 2016-02-10 四川中亚联邦科技有限公司 A kind of large data processing method in high in the clouds and system
CN106844089B (en) * 2015-12-03 2021-01-01 阿里巴巴集团控股有限公司 Method and equipment for recovering tree data storage
CN105868336B (en) * 2016-03-28 2019-11-29 哈尔滨工程大学 Spatial key word querying method in road network towards set
CN106067029B (en) * 2016-05-24 2019-06-18 哈尔滨工程大学 The entity classification method in data-oriented space
CN113297138A (en) * 2019-06-11 2021-08-24 武汉深之度科技有限公司 Index establishing method, data query method and computing device

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101484895A (en) * 2006-07-07 2009-07-15 新叶股份有限公司 Bit sequence search device, search method, and program
CN102043852A (en) * 2010-12-22 2011-05-04 东北大学 Path information based extensible markup language (XML) ancestor-descendant indexing method
US8234267B2 (en) * 2006-08-25 2012-07-31 Teradata Us, Inc. Hardware accelerated reconfigurable processor for accelerating database operations and queries

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101484895A (en) * 2006-07-07 2009-07-15 新叶股份有限公司 Bit sequence search device, search method, and program
US8234267B2 (en) * 2006-08-25 2012-07-31 Teradata Us, Inc. Hardware accelerated reconfigurable processor for accelerating database operations and queries
CN102043852A (en) * 2010-12-22 2011-05-04 东北大学 Path information based extensible markup language (XML) ancestor-descendant indexing method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于iMeMex数据模型的数据空间索引方法研究;王红滨等;《中国期刊全文数据库 计算机科学与探索》;20140131;第8卷(第1期);全文 *

Also Published As

Publication number Publication date
CN103902699A (en) 2014-07-02

Similar Documents

Publication Publication Date Title
CN103902699B (en) Data space retrieval method applied to big data environments and supporting multi-format feature
CN101359331B (en) Method and system for reordering search result
Chatzopoulou et al. The QueRIE system for Personalized Query Recommendations.
Harth et al. SWSE: Answers before links!
US20110307504A1 (en) Combining attribute refinements and textual queries
CN107784102A (en) A kind of data difference comparative approach based on oracle database
Salles et al. Intensional associations in dataspaces
CN102214216B (en) Aggregation summarization method for keyword search result of hierarchical relation data
Kanoje et al. Using MongoDB for social networking website deciphering the pros and cons
CN105022797A (en) Resource topic processing method and apparatus
KR101055306B1 (en) Web service based content management system
US20120310919A1 (en) Apparatus and method of searching for instance path based on ontology schema
CN104391952A (en) File system index establishing method and file system query implementing method
Kanoje et al. Using mongodb for social networking website
Amer-Yahia et al. SOCLE: Towards a framework for data preparation in social applications.
Chen et al. RAISE: a whole process modeling method for unstructured data management
Fogarolli et al. Agris-from a bibliographical database to a web data service on agricultural research information
Kotidis et al. Circumventing data quality problems using multiple join paths.
Tajima Schemaless Semistructured Data Revisited: —Reinventing Peter Buneman’s Deterministic Semistructured Data Model—
Yao et al. Efficient computation of multiple XML keyword queries
CN111104560B (en) Resource retrieval method and computer-readable storage medium
El Handri et al. Tweet Recommendation System Based on TA Algorithm and Natural Language Processing
Yang et al. A Novel PIM System and its Effective Storage Compression Scheme.
Charan et al. Addressing cold start problem in recommendation system using custom built hadoop ecosystem
Nguyen et al. An Architecture for Efficient RDF Data Management Using Structure Index with Relation-Based Data Partitioning Approach

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20170412