CN102043802B - Method for searching XML (Extensive Makeup Language) key words based on structural abstract - Google Patents
Method for searching XML (Extensive Makeup Language) key words based on structural abstract Download PDFInfo
- Publication number
- CN102043802B CN102043802B CN 200910197333 CN200910197333A CN102043802B CN 102043802 B CN102043802 B CN 102043802B CN 200910197333 CN200910197333 CN 200910197333 CN 200910197333 A CN200910197333 A CN 200910197333A CN 102043802 B CN102043802 B CN 102043802B
- Authority
- CN
- China
- Prior art keywords
- node
- tree
- coding
- xml
- structural
- 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.)
- Active
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a method for searching XML (Extensive Makeup Language) key words, comprising the following steps of: a, performing deep prior traversal to an XML tree, establishing a structural abstract tree of the XML tree, and respectively coding all the nodes of the structural abstract tree and all the leaf nodes of the XML tree so as to obtain the code of each leaf node and the code ofeach node of the structural abstract tree; b, for each leaf node and each node of the structural abstract tree, with the name of each node or a text value as a key and the code of each node as a value, establishing an inverted index; c, calculating the minimum and lowest public ancestor node from the structural abstract tree; and d, structuring searching return results on the basis of the calculated minimum and lowest public ancestor node.
Description
Copyright statement
Comprise content protected by copyright in the present patent application file; the copyright owner is had no objection to reprinting of carrying out of this content after openly by China national Department of Intellectual Property in present patent application to other unit or individual, but keeps all authority of other behavior opinion copyright of carrying out at this content with regard to other unit or individual.
Technical field
The present invention relates to key search, relate in particular to the XML key search.
Background technology
Key search is widely used in information retrieval field because being easy to use.Use key search, the user need not grasp the complicated query language, also need not to understand the pattern of XML document, just can find their interested content.The information of carrying out between structure webpage and enterprise is described and the standard of exchange because XML has become, the key search of carrying out at XML document (calling the XML keyword retrieval in the following text) thereby have extremely wide application prospect.
At present, the major part research relevant with the XML key search is all based on last common ancestor (LCA), wherein the most representative is by Yu Xu, the minimum last common ancestor (SLCA) that proposes among the XKSearch of Yannis Papakonstantinou in Efficient Keyword Search for Smallest LCAs inXML Databases one literary composition that SIGMOD ' 2005 delivers, wherein, be that the subtree that root node was returned is considered to significant to the user with the SLCA node.Among the XKSearch based on the SLCA node is calculated in the calculating of LCA.The easiest computing method of LCA are that the element in the document and text node are carried out the Dewey coding, and the result that Dewey is encoded calculates minimum common prefix and obtains LCA then.XRANK, MLCA, XSEarch, VLCA etc. are similar with SLCA in essence, repeat no more.XSeek, MaxMatch have then inquired into how to organize return results on the basis of SLCA.About XKSearch, can see for details
Http:// portal.acm.org/citation.cfm? id=1066217
Summary of the invention
Applicant of the present invention recognizes, uses key word to come XML is retrieved, and has following critical problem to need to solve:
1. the semanteme of definition of keywords retrieval how exactly
In view of the essence of the fuzzy search of key word, it is very important that the definition of its semanteme seems.
2. how to set up suitable search mechanism efficiently
Exist many redundancies and duplicate message in the index that the coded system of Dewey coding has determined document is set up, make that the index of document is also more many greatly than document usually.Fig. 1 a shows the example that an XML fragment is carried out the Dewey coding.Based on the Dewey coding, can determine two internodal mutual relationships easily and intuitively, and can determine the position of node in tree rapidly.The defective of Dewey coding is to have more redundant information, for example, the Dewey coding of leaf node Bob (0.0.0.0) in the XML tree shown in Fig. 1 (a) has comprised all routing informations of this node, and author (0.0.0) has repeated the coding to this path.
3. how to allow retrieval can return suitable result
At present, for the semanteme of SLCA node as key search, the result who returns is difficult to effective tissue.
Based on the understanding to above-mentioned technical matters, in the present invention, a kind of XML keyword search method based on structural summary has been proposed, it specifically comprises following aspect:
1) provide XKSS index establishing method, integrated structure summary technology is optimized the index size of XML key search;
2) provide the method for calculating SLCA on XKSS index basis;
3) provide and utilize XKSS to construct the minimum method that is communicated with tree.
Wherein, structural summary (structure summary) is a kind of approximate description to the XML structure that is obtained by the XML data, it is based on the routing information of node in the XML tree construction, yojan is carried out in node and path to the XML document tree, is applied to usually coupling is searched in the path of inquiries such as xPath fast.Its feature is that node and the path of all repetitions only occurs once in this structural summary, in other words, can not have two nodes with same paths in structural summary.For an XML document, structural summary is very little, and for example, the structural summary of the XML document shown in Fig. 1 a is shown in Fig. 1 b.
The inquiry of XML small-branch mode is the inquiry that has at the selection predicate of XML document structure and content in essence.It is the core operation that the XML flow data is handled that one group of small-branch mode and the XML document that arrives are at any time mated.
Many relevant researchs are also being arranged aspect the structural summary index, and wherein, DataGuide is suggested the earliest and is applied in the Lore project of Stanford university.The basic thought of DataGuide is that the method that NFA is converted to DFA is applied in the same paths on the reduction XML tree.DataGuide is a more rough structural summary, wherein may comprise in the real data and non-existent routing information, so, some had been proposed afterwards successively based on two structural summaries like (bisimilar) equivalent concepts again, as 1-index and xSKETCH, and loosened afterwards two like condition obtain the k rank two like A (k)-indexes.
According to an aspect of the present invention, provide a kind of method that can realize the XML key search efficiently, wherein, the XML index for the XML key search provided by the present invention is designated as XKSS, and said method may further comprise the steps:
A. the method for traditional use Dewey coding being set up index is expanded, and leaf node and the structure node of XML tree are separately encoded;
B. the data after will encoding are according to value set up inverted index respectively with label;
C. based on the XKSS index, the computational algorithm of SLCA is improved, thereby utilize the XKSS index calculation to go out the SLCA node.
Description of drawings
By the description of at least one indefiniteness embodiment of the present invention being done below in conjunction with accompanying drawing, other characteristics of the present invention, advantage will seem more clear and obvious.Wherein:
Fig. 1 a shows the example that an XML fragment is carried out the Dewey coding;
Fig. 1 b shows the structural summary of XML document shown in Fig. 1 a;
Fig. 2 a-2b shows structural summary tree and the coding that a specific embodiment according to the present invention is set up;
Fig. 3 a-3c shows a process of setting up minimum connection tree according to SLCA;
Fig. 4 shows the present invention and the experiment comparative result of prior art on the index size;
Fig. 5 shows the present invention and the experiment comparative result of prior art on algorithm execution time, and wherein, label contains key word;
Fig. 6 shows the present invention and the experiment comparative result of prior art on algorithm execution time, and wherein, label does not contain key word.
In the accompanying drawings, same or analogous Reference numeral is represented same or analogous technical characterictic.
Embodiment
Come the present invention is described in detail below with reference to instantiation, hereinafter be divided into two parts generally, first relates generally to the semantical definition of XML key search, and foundation and SLCA on this basis that second portion then relates to above-mentioned XKSS index calculate.
One, the semanteme of XML key search
Before address, a major issue of XML key search is to determine the semanteme of key search, determines that namely the user is interested in which part of XML.Like this, can avoid returning information too much and that key word is irrelevant.Current research mainly is the semanteme that the XML key search is determined on the basis with LCA, and what wherein the most generally adopt is the SLCA concept that XKSearch proposes, and the specific definition of these concepts will become clearer by following description.
Usually, XML document can be represented as the model of tree, and for an XML tree T, each node v wherein can be corresponding to the element in the XML document.Usually, suppose that text node only appears in the leaf node, the description of file structure is then appeared at non-leaf node usually, also claim semantic node.For convenience of description, (tag name of v) representing node v, (textual value of v) representing node v, Depth (v) represents the degree of depth of v in tree T to Text to use Label herein.If use the Dewey coded representation, the length of the Dewey of node coding is exactly the degree of depth of this node.If the node v in tree T is the ancestors of w, then be expressed as v＜w;
The LCA of two nodes refers to the node of degree of depth maximum among the public ancestors of these two nodes.
Given two node a, the Dewey coding of b, the maximum-prefix of these two Dewey codings is exactly a, the Dewey coding of the LCA node of b.
Given a series of key word w
1, w
2... w
k, for each key word w wherein
i, in XML tree T, can find a node set S of correspondence with it
i, this node set S
iIn each node comprise w
i
Claim LCA (S herein,
1, S
2... S
k) be the LCA node set of key word.
Equally, for node set S
1, S
2... S
k, when following condition that and if only if satisfies simultaneously, v ∈ SLCA (S
1, S
2... S
k):
1.V∈LCA(S
1，S
2...S
k)；
2. do not exist any w to satisfy w ∈ LCA (S
1, S
2... S
k); And
3.v＜w。
Given a series of key word w
1, w
2... w
kWith XML tree T, w
1, w
2... w
kReply the subtree that comprises each at least one example of key word that subtree refers to T.w
1, w
2... w
kMinimum reply subtree and meet the following conditions:
1. itself be one and reply subtree;
2. in each subtree of this subtree self, neither one is to reply subtree.
Given a series of key word w
1, w
2... w
kWith XML tree T, w
1, w
2... w
kMinimum be communicated with tree and refer to (the w with SLCA
1, w
2... w
k) be root node, only comprise from the root node to w
1, w
2... w
kThe path on the tree of all nodes.
Two, the calculating of the foundation of XKSS index and SLCA
In the XML of reality database, what the index amount usually can be very is huge.And XKSS is by introducing the concept of structural summary, employing is encoded separately to the leaf node of XML document, other nodes use the mode of structural summaries tree that its structure is encoded, significantly yojan identical routing information among the XML, reduced the number of nodes that needs coding.So the coding of entire document tree is divided into two parts: text node coding and structural summary tree-encoding.Again two kinds of codings are set up index respectively at last.This method main cause of compressed index size significantly replaces the coding of all the non-leaf nodes in the document with regard to being the coded message of using structural summary to set, and the size of structural summary tree is to be far smaller than XML tree itself.
The foundation of XKSS index and use are carried out two stages respectively: (run-time) stage when pre-service (pre-process) stage and operation.
Pretreatment stage, by XML document is carried out depth-first traversal, use the mode of Dewey coding that node is encoded, set up the structural summary tree of document simultaneously, the structural summary tree is also used Dewey coded representation, two parts to be coded in once and finishes simultaneously in the traversal.Dewey coding to the node in the leaf text node in the document and the structural summary tree is that key is set up inverted index respectively with the nodal value at last.
Stage during operation, the user submits keyword query to, finds out the node that comprises key word by inverted index, utilizes the SLCA algorithm to calculate the Dewey encoded radio of SLCA node, because the just leaf node of index, so can not obtain real SLCA node according to the Dewey coding of SLCA.At this time can utilize Dewey coding and the structural summary index of SLCA to find out actual SLCA node.
XKSS carries out index to the structural summary tree of the text node in the XML document tree and the document respectively.The foundation of structural summary need travel through XML document, can utilize the structural summary that XML document is carried out setting up when Dewey encodes XML document.
The structural summary tree of setting up at last and coding are shown in Fig. 2 a-2b.Wherein, Fig. 2 a is the coding to text node, only the node that comprises textual value is kept coding and index during index.The coding of each node v adopts that (wherein, Dcode is the Dewey coding of node v in the XML tree for Dcode, form id), and the sign of the parent node of id representation node v can find the position of the parent node of v in the structural summary tree by this sign.Fig. 2 b then is that the structural summary of document among Fig. 2 a is set, and uses the Dewey coding too, and different is that the root node that structural summary is set is encoded as 1, so the coding of all nodes is all with 1 beginning, thus the differentiation of the coding of realization and former document tree.The node that marks with round dot among the figure is the parent node of text node, same with (Dcode, id) form coding, wherein, Dcode is the Dewey coding of this node in the structural summary tree, and id represents a sign of this node, and this sign is corresponding with the sign of document tree Chinese version node, by this sign, text node can find the parent node of oneself.
It is very directly perceived that XML document is carried out the Dewey Methods for Coding, and the method for setting up the structural summary tree can be with reference to the algorithm of setting up DataGuide.Be that document is carried out the Dewey coding or set up the structural summary tree all needing document is once traveled through, finish so we are combined in these two processes in a document traversal.Finished after the coding to whole XML tree and this structural summary tree, be key to the node in the text leaf node in the XML tree and the structural summary tree with nodename or textual value then, the Dewey value of being encoded to is set up inverted index, and the coding of all the non-leaf nodes in the XML tree then can not carry out index.Use the node in the structural summary tree to replace all the non-leaf nodes in the original XML tree to carry out index, reduced the number of nodes that carries out index greatly, this also is that the XKSS index is much smaller than the major reason than primary index method.
The algorithm of the above-mentioned XKSS of foundation index can also be expressed as:
The foundation of algorithm 1:XKSS index
Use structural summary that index is compressed, can effectively reduce the size of index.Yet the structural summary tree has replaced the structural information of entire document in the index, thus must from structural summary, calculate the SLCA node, and construct final return results.
The key word that the user proposes may be to the inquiry of text node Bob for example, and what also may represent is that the user wishes for example author of the semanteme inquired about.For text node, complete index has been arranged, can directly adopt the SLCA algorithm to calculate; And for semantic node, owing to only kept its structural information in this example, therefore in one embodiment of the invention, the SLCA computational algorithm based on XKSS having been proposed, experiment has also proved this algorithm validity.
At first, given key word w
1, w
2... w
k, w
K+1, w
K+2... w
j, in this example, key word is divided into two classes, key word w
1, w
2... w
kAppearing in the text node, is the inquiry to text node, key word w
K+1, w
K+2... w
jAppearing in the semantic node, is to inquiring about semantic replenishing.
In this example, separately handled between above-mentioned two class key words, for w
1, w
2... w
k, from index, find the set of node S that comprises these key words
1, S
2... S
k, calculate all S by the SLCA algorithm then
1, S
2... S
kThe Dewey coding of SLCA node, these Dewey codings can not find corresponding node in index, because intermediate node can't be indexed, so need utilize structural summary to find real SLCA node.
Function given below can be obtained its coding and title in the structural summary tree by the Dewey coding of a SLCA node.If key word does not appear in the semantic node, then SLCA has just been calculated.If there is key word to appear in the semantic node, so also need further to calculate, conclude above-mentioned situation, algorithm SSB-SLCA has been proposed, below also will obtain describing in detail.
The algorithm of above-mentioned calculating SLCA can also be expressed as (the whole following one page that moves into this instructions of the description of this algorithm):
Algorithm 2: based on XKSS index calculation SLCA
If replying subtree with minimum is the result, then may return whole XML document in the worst case, wherein be root node with SLCA.In this example, utilize structural summary to obtain setting as return results minimum the connection if discuss.
Can be obtained the Dewey coding of SLCA set of node in the structural summary tree of key word by before work, the minimum of utilizing this Dewey coding can adopt the bottom-up algorithm construction to go out key word is communicated with subtree.
Algorithm 3: use minimum connection of XKSS index structure and set
Input: SLCA node v ' and comprise the node v of key word
1, v
2... v
k
Export: be a minimum tree T that is communicated with of root node with v '
a
Step 1: with v
1, v
2... v
kAdd T as leaf node
a
Step 2: for v
1, v
2... v
kIn each node v
i, find that id is v in the structural summary tree
i.id node v
i', as v
iParent node add T
a, in the structural summary tree, find v equally
i' parent node, and it is added T
aIn as v
i' at T
aIn parent node, add T up to SLCA node v '
a
Step 3: the Dewey according to each vi is encoded to information with T
aIn node split into and satisfy each v
iThe routing information of Dewey coding.
The implementation of algorithm is described with an example equally.Proposing key word equally in XML document shown in Fig. 2 a is " James, Editorial, Bob, SignalML ", passes through the calculating of algorithm 2 so, will obtain such conclusion: the SLCA node is dblp (0) node among Fig. 2 b.Execution algorithm 3, T
aWill process as shown in Figure 3 be established.
At first carry out above-mentioned steps 1, obtain the result of Fig. 3 a, all leaf nodes that comprise key word are added into T
aCarry out above-mentioned steps 2 then, according to the structure of structural summary tree, bottom-up node is added T
aBecause the node in the structural summary tree is through merging, thus also need according to the Dewey coding of leaf node wherein actual node launch.For example the front two of the Dewey of the Bob among Fig. 3 b and SignalML coding is 0.0, and the front two of the Dewey of James and Editorial coding is 0.1000, this explanation is at the 2nd layer of tree, Bob and SignalML have same ancestors, and James and Editorial have same ancestors, so the 2nd layer article node is divided into two in will setting, respectively as Bob, SignalML and James, the ancestor node of Editorial.
By carrying out above-mentioned steps 3, the result shown in Fig. 3 c will be obtained.
In order to prove the present invention than all advantages of prior art, the applicant has carried out experiment relatively especially, and concrete outcome is shown in Fig. 4-6.
More than specific embodiments of the invention are described.It will be appreciated that the present invention is not limited to above-mentioned specific implementations, those skilled in the art can make various distortion or modification within the scope of the appended claims.
Claims (4)
1. XML keyword search method may further comprise the steps:
A. an XML tree is carried out depth-first traversal, set up the structural summary tree of described XML tree, and all nodes of described structural summary tree and all leaf nodes of described XML tree are encoded respectively, with the coding that obtains each leaf node and the coding of each structural summary tree node;
B. to described each leaf node and each structural summary tree node, be key with its nodename or textual value separately, with its value of being encoded to separately, set up inverted index;
C. by calculating minimum last common ancestor node in the described structural summary tree;
D. based on the minimum last common ancestor node that calculates, construct the retrieval return results;
Wherein, described steps d comprises:
-with minimum last common ancestor node v ' with comprise the node v of key word
1, v
2V
kAs input, obtaining one is the minimum connection tree T of root node with v '
a, as output,
Wherein, the user submits keyword query to, finds out the node that comprises key word by inverted index, and utilizes structural summary to obtain setting as return results minimum the connection.
2. method according to claim 1, wherein, described step a comprises:
-by a degree of depth traversal, use the Dewey coding that the leaf node of described XML tree is encoded, and set up described structural summary tree, and set with the described structural summary of Dewey coded representation, thereby the coding of each leaf node and the coding of each structural summary tree node obtained.
3. method according to claim 1, wherein, described step c comprises:
-find out the node that comprises the key word that the user provides;
-utilize minimum last common ancestor node algorithm to calculate the Dewey coding of minimum last common ancestor node;
-utilize Dewey coding and the structural summary index of minimum last common ancestor node to find out the SLCA node.
4. method according to claim 1, wherein, described steps d comprises:
D1. with v
1, v
2V
kAdd T as leaf node
a
D2. for v
1, v
2V
kIn each node v
i, in structural summary tree, find the v that is designated of a parent node
i.id node v
i', with v
i' add T as the parent node of vi
a, and, in the structural summary tree, find v
i' parent node, and it is added T
aIn as v
i' at T
aIn parent node, be added into T up to v '
a
D3. according to each v
iDewey coding, with T
aIn node be split as and satisfy each v
iThe routing information of Dewey coding.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910197333 CN102043802B (en) | 2009-10-16 | 2009-10-16 | Method for searching XML (Extensive Makeup Language) key words based on structural abstract |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910197333 CN102043802B (en) | 2009-10-16 | 2009-10-16 | Method for searching XML (Extensive Makeup Language) key words based on structural abstract |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102043802A CN102043802A (en) | 2011-05-04 |
| CN102043802B true CN102043802B (en) | 2013-09-25 |
Family
ID=43909944
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200910197333 Active CN102043802B (en) | 2009-10-16 | 2009-10-16 | Method for searching XML (Extensive Makeup Language) key words based on structural abstract |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102043802B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102867054A (en) * | 2012-09-13 | 2013-01-09 | 江苏乐买到网络科技有限公司 | XML (extensible markup language) keyword query method |
| CN111339457B (en) * | 2018-12-18 | 2023-09-08 | 富士通株式会社 | Method and apparatus for extracting information from web page and storage medium |
| CN110245248B (en) * | 2019-05-31 | 2022-10-21 | 河南大学 | Remote sensing image keyword query method |
| CN112749301B (en) * | 2020-10-12 | 2022-09-30 | 河南大学 | Keyword query method for fuzzy XML (extensive makeup language) of massive remote sensing metadata |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1585945A (en) * | 2001-09-28 | 2005-02-23 | 甲骨文国际公司 | Mechanism for mapping XML schemas to object-relational database systems |
| CN101091181A (en) * | 2004-10-05 | 2007-12-19 | 尤尼西斯公司 | Saving and restoring an interlocking trees datastore |
-
2009
- 2009-10-16 CN CN 200910197333 patent/CN102043802B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1585945A (en) * | 2001-09-28 | 2005-02-23 | 甲骨文国际公司 | Mechanism for mapping XML schemas to object-relational database systems |
| CN101091181A (en) * | 2004-10-05 | 2007-12-19 | 尤尼西斯公司 | Saving and restoring an interlocking trees datastore |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102043802A (en) | 2011-05-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP4516566B2 (en) | Retention of hierarchy information in mapping between XML document and relational data | |
| Fegaras et al. | Query processing of streamed XML data | |
| Holland et al. | Choosing a data model and query language for provenance | |
| US20040128296A1 (en) | Method for storing XML documents in a relational database system while exploiting XML schema | |
| US20060206466A1 (en) | Evaluating relevance of results in a semi-structured data-base system | |
| CN107992608B (en) | SPARQL query statement automatic generation method based on keyword context | |
| JP2009518718A (en) | Simple index structure for XML | |
| CN102043802B (en) | Method for searching XML (Extensive Makeup Language) key words based on structural abstract | |
| CN100397397C (en) | XML data storage and access method based on relational database | |
| CN101615190B (en) | Safe XML keyword search method | |
| CN101833588B (en) | Indexing structure of XML (Extensive Markup Language) document | |
| CN103294791A (en) | Extensible markup language pattern matching method | |
| US20090307187A1 (en) | Tree automata based methods for obtaining answers to queries of semi-structured data stored in a database environment | |
| CN102033956A (en) | Graphical XML content and structure query system with intelligent prompt function | |
| CN1560763B (en) | Method for translating expandable mark language path inquiry into structure inquiry | |
| CN1564161A (en) | Method of constituting and browsing semantic links | |
| Qin et al. | Efficient XML query and update processing using a novel prime-based middle fraction labeling scheme | |
| Raji et al. | RDF approach on social network analysis | |
| KR20080008573A (en) | Method for extracting association rule from xml data | |
| Zhang et al. | Odaies: ontology-driven adaptive Web information extraction system | |
| CN102867054A (en) | XML (extensible markup language) keyword query method | |
| Kung et al. | A novel twig-join swift using SST-based representation for efficient retrieval of internet XML | |
| Álvarez-García et al. | Graphgen: A tool for automatic generation of multipartite graphs from arbitrary data | |
| Chavez et al. | Rdf-based model for encoding document hierarchies | |
| Min et al. | Method of Understanding Structure and Building Database with Material Experiment Data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |