CN105447188B - A kind of reciprocity social networks document retrieval method of knowledge based study - Google Patents
A kind of reciprocity social networks document retrieval method of knowledge based study Download PDFInfo
- Publication number
- CN105447188B CN105447188B CN201510955213.4A CN201510955213A CN105447188B CN 105447188 B CN105447188 B CN 105447188B CN 201510955213 A CN201510955213 A CN 201510955213A CN 105447188 B CN105447188 B CN 105447188B
- Authority
- CN
- China
- Prior art keywords
- node
- interest
- list
- knowledge
- recommended
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 239000013598 vector Substances 0.000 claims abstract description 32
- 230000008569 process Effects 0.000 claims abstract description 8
- 230000003044 adaptive effect Effects 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 6
- 230000001105 regulatory effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 description 9
- 230000008859 change Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000006855 networking Effects 0.000 description 2
- 241000404883 Pisa Species 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000005538 encapsulation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000011273 social behavior Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of knowledge baseds to learn reciprocity social networks document retrieval method, it belongs to social networks technical field, mainly solves the problems, such as that file retrieval recall rate is low, forwarding expense is big and performance is not satisfactory in reciprocity social networks.Its step is：A) node establishes interest index and knowledge index；B) node obtains recommended node forwarding inquiries message by interest index and knowledge index.Node establishes interest index according to the number of documents with the acquisition of the interest vector keyword of its interest same node point, and establishes knowledge index according to the number of documents that searching keyword obtains.In forwarding inquiries message, recommended node is obtained from interest index and knowledge index, message is transmitted to recommended node.This method learning knowledge in query process establishes index, and obtains preferable inquiry according to index, compared with some existing typical methods, achieves higher recall rate, lower network overhead and preferable performance.
Description
Technical field
The present invention relates to social networks technical fields, and knowledge learning equity social networks document proposed by the present invention is examined
Suo Fangfa (IESLP) is used for the file retrieval of reciprocity social networking service.
Background technology
Social networking application is more and more extensive, and the social activity of virtual community may be implemented by social networks by people, such as
Friend-making, mutual assistance, publication commercial advertisement, carries out resource sharing and retrieval etc. at chat.Social networks has there are many type based on visitor
The online social networks such as FaceBook, Renren Network etc. of family machine/server mode；There is the mobile social activity based on honeycomb soverlay technique
Network, such as wechat；Also there is the reciprocity social networks based on P2P technologies.
P2P technologies are also referred to as peer-to-peer, and the node in reciprocity social networks is both server and client.?
In the social networks built using non-structural P 2 P technology, search file is a challenging project.Have one at present
A little methods can be used for the file retrieval of reciprocity social networks, such as (1) random breadth-first search technology (RBFS)
[V.Kalogeraki,D.Gunopulos,D.Zeinalipour-yazti.A Local Search Mechanism for
Peer-to-Peer Networks[C].Proc.Of the 11th ACM Conference on Information and
Knowledge Management(CIKM'02).New York:ACM,2002:300-307.].(2) NeuroGrid technologies
[Joseph S.NeuroGrid:Semantically routing queries in peer-to-peer
networks.Proceedings of the International Workshopon Peer-to-Peer Computing,
Pisa,Italy,2002.].(3) ESLP technologies [L.Liu, N.Antonopoulos, S.Mackin, J.Xu, D.Russell,
Efficient Resource Discovery in Self-organized Unstructured Peer-to-Peer
Networks,Concurrency and Computation:Practice and Experience,Wiley,Vol 23(2),
February 2009, pp.159-183.] etc..RBFS methods randomly choose k neighbor node when carrying out query messages processing
Query messages forwarding is carried out, the neighbor node for receiving message randomly chooses k neighbours' forwarding inquiries message again, until TTL exhausts,
The method efficiency of this search file is low, be delayed length.NeuroGrid technologies establish knowledge base in network node, will inquire
In the knowledge store to knowledge base learnt in journey, message forwarding, forwarding are carried out according to the knowledge-chosen recommended node in knowledge base
Number of nodes is between minimum forwarding degree and max-forwards degree.This method ratio RBFS has improvement, and recall rate and network performance are all
It has been improved.ESLP is a kind of novel reciprocity social network search technology, and the interpersonal relationships during it associates people is managed
By applying in file retrieval, the Social behaviors Fast Learning knowledge of people is simulated, social circle is spontaneously formed, improves file retrieval
Success rate, this method have larger improvement in aspect of performance ratio NeuroGrid.
Although the file retrieval of reciprocity social networks may be implemented in these existing methods, but there is also some defects.
The forwarding degree of RBFS is a fixed constant；The forwarding degree of NeuroGrid between minimum forwarding degree and max-forwards degree,
But it is unable to adaptive change；The forwarding degree of ESLP although it is contemplated that destination node and searching keyword correlation adaptive change,
But destination node number of documents and the relationship of searching keyword are not accounted for, and does not account for minimum forwarding degree and most yet
The adaptive change of big forwarding degree.User forms community by interest in social networks, these algorithms also use by not explicit excavation
The interest attribute at family, to which by the comparison self-teaching of interest vector similitude, quick clustering is at community.It is existing that these are right
Equal social networks file retrieval technical method also has the space promoted in recall rate, forwarding expense, network performance etc..
Invention content
In order to solve above-mentioned reciprocity social network environment, hereafter document search recall rate is relatively low, network overhead is big, performance is not high
Defect, the present invention proposes a kind of reciprocity social networks document retrieval method (abbreviation IESLP) of knowledge based study, network section
Point is according to interest vector and searching keyword self-teaching, automatic Community Formation cluster, to improve under reciprocity social network environment
File retrieval recall rate and comprehensive performance.Using each node in the reciprocity social networks of IESLP document retrieval methods
Effect is equal, and node executes file retrieval routing algorithm and is divided to two kinds of situations：It generates and sends when the querying node document
Initial query message, and interest index and knowledge index are establishd or updated according to the feedback message of other nodes in network；When this
By the local number of documents with Keywords matching of statistics when node receives the query messages from other nodes, sent out to query node
Feedback message is sent, and selects the neighbor node forwarding inquiries message of oneself.Realize that technical scheme of the present invention includes as follows：
A kind of reciprocity social networks document retrieval method of knowledge based study, includes the following steps：
Step A, node establish interest index and knowledge index, including：During file retrieval, node is identical from interest
Destination node obtain the similar knowledge store of interest in local interest concordance list, while obtaining and looking into according to searching keyword
The knowledge store of Keywords matching is ask in Indigenous knowledge concordance list；
Step B, node obtain neighbor node by local interest index and knowledge index and are used as recommended node, and to recommendation
Node forwarding inquiries message, including：When node needs the query messages for forwarding other nodes to send over, local interest rope is inquired
Draw table and knowledge index table obtains the neighboring node list for including matching inquiry keyword document, is calculated according to matching number of documents
The correlation coefficient of neighbor node in list, and minimum forwarding degree and max-forwards degree is combined to calculate adaptive forwarding degree, then
Adaptive forwarding degree selection recommended node according to neighbor node in list carries out query messages forwarding.
As optimal technical scheme, the step A establishes interest index and the process of knowledge index is as follows：
Step 1), when query node is identical with destination node interest, interest of the query node according to destination node feedback
Vectorial keyword and destination node include that the number of documents of interest vector keyword establishs or updates local interest concordance list；
Step 2), when query node and destination node interest difference or query node is identical with destination node interest
And destination node interest vector lists of keywords do not include searching keyword when, query node according to destination node feedback inquiry
Keyword and destination node include that the number of documents of searching keyword establishs or updates Indigenous knowledge concordance list；
Step 3), destination node is according to searching keyword statistical match number of documents.
As optimal technical scheme, it is using neighbor node as the determination method of recommended node in the step B：If list
The adaptive forwarding number of degrees value of middle neighbor node is more than the quantity for having selected recommended node, then the neighbor node is chosen as recommending section
Point.
As optimal technical scheme, the process that the step B selections recommended node carries out query messages forwarding is as follows：
Step 1) is in the interest vector that searching keyword is included in node, then according to searching keyword from interest concordance list
Middle selection node is added in recommended node list, when recommendation list interior joint lazy weight max-forwards number of degrees value, then foundation
Searching keyword selects node to be added in recommended node list from knowledge index table, when recommendation list interior joint lazy weight
Minimum forwarding number of degrees value is then added to from the remaining section of random selection in neighbor list in recommended node list；
Step 2) is not when inquiry subject key words are in the interest vector of present node, then according to searching keyword from knowledge
In concordance list select node be added in recommended node list, when recommendation list interior joint lazy weight minimum forward number of degrees value,
Then it is added in recommended node list from the remaining section of random selection in neighbor list；
Step 3) when not meeting the recommended node of search request in interest concordance list and knowledge index table, search by expansion
Rope range increases the numerical value of minimum forwarding degree；Random selection meets newest minimum turn from the connection neighbor node of present node
The node that hair degree requires is added to as recommended node in recommended node list；
4) node forwarding inquiries message is selected from recommended node list successively, until list is empty for recommended node.
As optimal technical scheme, correlation coefficient is calculated in the step B, specially：
The correlation coefficient of i-th of the node obtained from interest concordance list or knowledge index table according to searching keyword isWherein i=1,2 ..., n, n are to be obtained from interest concordance list or knowledge index table according to searching keyword
Neighbor node number, the molecule m of item on the right of equationiFor i-th of node and the matched number of documents of searching keyword, denominatorFor the matching number of documents of node that is obtained from interest concordance list or knowledge index table according to searching keyword and.
As optimal technical scheme, the value that adaptively forwarding degree is calculated in the step B is：
The adaptive forwarding degree k of i-th of nodei=Round (ri λ(dmax-dmin))+dmin, wherein i=1,2 ..., n, ri
For the correlation coefficient of i-th of node, dminIt is for minimum forwarding degree, i.e., at least a by the neighbor node of selection forwarding inquiries message
Number, dmaxFor max-forwards degree, the i.e. maximum node number by selection forwarding inquiries message, λ is index regulatory factor, and range is 0
Between~1, Round functions are bracket function.
As optimal technical scheme, the minimum forwarding degree dmin=2, max-forwards degree dmax=3, index regulatory factor λ
=0.7.
As optimal technical scheme, the interest concordance list and the knowledge index table are the identical Hash table knot of structure
Structure, keyword constitute the key of Hash table, including the neighboring node list of matching inquiry keyword document constitutes the cryptographic Hash of key.
As optimal technical scheme, the element in the list includes two domains, and a domain stores information of neighbor nodes, separately
One domain stores the number of documents that the neighbor node includes corresponding Keywords matching；It include information of neighbor nodes in the list
Element by number of files magnitude Bit-reversed from big to small, randomly selected node acquiescence matching number of files magnitude is 0, comes row
Table end.Compared with prior art, beneficial effects of the present invention：
The present invention indexes and knowledge index in reciprocity social networks joint structure interest, during file retrieval, node
Same interest information of neighbor nodes is stored by number of files magnitude in oneself interest concordance list from big to small, and will be closed by inquiry
The nodal information that keyword retrieves is stored by number of documents in knowledge index table from big to small, when forwarding inquiries message, node
According to study to the more neighbor node of knowledge acquisition number of documents as forwarding recommended node.In this way, improving reciprocity social network
The recall rate and retrieval performance of network file retrieval.
Description of the drawings
Fig. 1 is interest concordance list and knowledge index table structure chart；
Fig. 2 is the implementation illustration for establishing interest index and knowledge index；
Fig. 3 is the implementation illustration that node obtains recommended node forwarding inquiries message by interest index and knowledge index；
Fig. 4 is the comparison of technical solution of the present invention IESLP and ESLP, NeuroGrid method recall rate；
Fig. 5 is the comparison of technical solution of the present invention IESLP and ESLP, NeuroGrid method forwarding expense；
Fig. 6 is that technical solution of the present invention IESLP often forwards a query messages to be called together with ESLP, NeuroGrid method
The comparison for the rate of returning.
Specific implementation mode
Further description is done to technical solution of the present invention below by the drawings and specific embodiments.
The realization of the present invention includes two steps A and B：
Step A：Node establishes the step of interest index and knowledge index；
When node (query node) inquires document, generates query messages and send query messages, inquiry to neighbor node
Node obtains the similar knowledge store of interest in local interest concordance list from the identical destination node of interest, while according to inquiry
Keyword obtains required knowledge store in Indigenous knowledge concordance list.The process for establishing interest index and knowledge index is as follows：
1) when query node is identical with destination node interest, query node is closed according to the interest vector of destination node feedback
Keyword and destination node include that the number of documents of interest vector keyword establishs or updates local interest concordance list；
2) when query node, different or query node is identical with destination node interest with destination node interest and target section
Point interest vector lists of keywords do not include searching keyword when, query node according to destination node feedback searching keyword and
Destination node includes that the number of documents of searching keyword establishs or updates Indigenous knowledge concordance list；
3) destination node is according to searching keyword statistical match number of documents.
Fig. 1 is the structure chart of interest concordance list and knowledge index table of the present invention.Interest concordance list and knowledge index table are
The identical Hash table structure of structure, keyword constitute the key of Hash table, the neighboring node list structure of matching inquiry keyword document
The cryptographic Hash of bonding；Element in list includes two domains, and a domain stores neighbor node address information, and the storage of another domain should
Neighbor node includes corresponding keyword (key) matched number of documents.The element comprising information of neighbor nodes presses document in list
Quantitative value Bit-reversed from big to small, randomly selected node acquiescence matching number of files magnitude is 0, comes list end.
In Fig. 1, left-hand line Topic_n in concordance listi(wherein i=1,2 ..., y, y are the quantity of key) is Hash table
Key is made of interest keyword or searching keyword；Right side is the corresponding ltsh chain table of key, the Node_n in chained list nodei(i
=1,2 ..., x) be query node neighbor node, mi(i=1,2 ..., x) it is the document that neighbor node includes keyword (key)
Quantity, and mi>mi+1, x is the quantity of ltsh chain table node.
Fig. 2 is the one embodiment for establishing interest index and knowledge index.In fig. 2, node A, B, D has same interest,
Interest vector (Interest Vector) isNode C, E interest is identical, they
Interest vector isNode A is interested in theme " iPhone ", generates inquiry and disappears
It ceases, encapsulation theme " iPhone ", query messages are broadcast to neighbor node B, C, D by A respectively in message.The interest of node B, D and A
Identical, they read keyword C#, Java and Python in interest vector respectively, search the local for including these three keywords
Document, and according to keyword statistic document quantity.Statistical information is fed back to A by B and D, and interest is written in the information received by node A
In concordance list (Interest Index).Including the document B of C# themes has 9, D has 6, therefore, right in interest concordance list
Answer B in the knot vector of C# keywords that should come before D.And to include the document B of Java keywords have 3, D has 8, therefore
D in the knot vector of Java keywords is corresponded in interest concordance list to be come before B.Include the document B of Python keywords
Each 6 with D, A stores information according to the sequencing for receiving feedback message.Assuming that in Fig. 2, A first receives the feedback of B, after receive D
Feedback, then B come before D.
The interest of node C and A is different, but has the interested themes of A " iPhone " in the document of the locals node C.Node C is at this
The document for including theme " iPhone " is searched on ground, is often found a document matches number met the requirements and is increased one.8 texts are found altogether
Shelves, node C successful search numbers are 8.C feeds back information to A, and node A deposits the information received with " iPhone " for keyword
It stores up in knowledge index table (Knowledge Index).C forwards messages to neighbor node E.Node E is looked into local document
The document 2 for including theme " iPhone " is found, then node E successful searches number is 2.At this point, successful search number in network
10 (8+2=10) will be increased.E feeds back information to A, in the knowledge index table of A, with the node that " iPhone " is keyword
The connection that after C has 8 documents, node A to receive the feedback information of E in vector, will be inserted in behind C, and establish to E nodes.
It is worth noting that although not including inquiry master in the interest vector of destination node identical with query node interest
Keyword is inscribed, but destination node there may be the interested document of query node, therefore, destination node is removed to be searched by interest
Outside, it should also be searched by searching keyword.In Fig. 2, it includes " iPhone " to have a document in node B, therefore the success of B node is looked into
It is 1 to look for number, and B node feedback information is inserted into the information storage to knowledge index received behind node E to A, A.
Step B：Node obtains neighbor node by local interest index and knowledge index and is used as recommended node, and to recommendation
The step of node forwarding inquiries message.
When node receives the query messages that other nodes send over, if necessary to forwarding inquiries message, then inquire
Local interest concordance list and knowledge index table obtain the neighboring node list for including matching inquiry keyword document, according to matching text
The correlation coefficient of neighbor node in gear number amount calculations list, and minimum forwarding degree and max-forwards degree is combined to calculate adaptive turn
Hair degree, the then adaptive forwarding degree selection recommended node progress query messages forwarding according to neighbor node in list.If list
The adaptive forwarding number of degrees value of middle neighbor node is more than the quantity for having selected recommended node, then the neighbor node is chosen as recommending section
Point.
According to the correlation coefficient of neighbor node in matching number of documents calculations list, specially：
The correlation coefficient of i-th of the node obtained from interest concordance list or knowledge index table according to searching keyword isWherein i=1,2 ..., n, the molecule m of equation the right itemiIt is matched with searching keyword for i-th of node
Number of documents, denominatorMatching for the node obtained from interest concordance list or knowledge index table according to searching keyword
Number of documents and the number for the neighbor node that n obtains for foundation searching keyword from interest concordance list or knowledge index table.
Calculating adaptive forwarding degree is specially：
The adaptive forwarding degree k of i-th of nodei=Round (ri λ(dmax-dmin))+dmin, wherein i=1,2 ..., n, ri
For the correlation coefficient of i-th of node, dminIt is for minimum forwarding degree, i.e., at least a by the neighbor node of selection forwarding inquiries message
Number, dmaxFor max-forwards degree, the i.e. maximum node number by selection forwarding inquiries message, λ is index regulatory factor, and range is 0
Between~1, Round functions are bracket function, and n is the neighbour obtained from interest concordance list or knowledge index table according to searching keyword
Occupy the number of node.
The process that the selection recommended node carries out query messages forwarding is as follows：
1) in the interest vector that searching keyword is included in node, then according to searching keyword from the local interest of node
Neighbor node of the selection comprising matching inquiry keyword document is added in recommended node list in concordance list；When recommended node arranges
Neighbor node quantity deficiency max-forwards number of degrees value in table is then selected according to searching keyword from the Indigenous knowledge concordance list of node
It selects the neighbor node comprising matching inquiry keyword document to be added in recommended node list, if in recommended node list at this time
Minimum forwarding number of degrees value that neighbor node quantity is insufficient, then randomly choose remaining neighbor node from the neighboring node list of node and add
It is added in recommended node list, until neighbor node quantity reaches minimum forwarding number of degrees value in recommended node list；
2) when inquiry subject key words are not in the interest vector of node, then foundation searching keyword is directly from the sheet of node
Neighbor node of the selection comprising matching inquiry keyword document is added in recommended node list in ground knowledge index table, works as recommendation
The insufficient minimum forwarding number of degrees value of neighbor node quantity, then randomly choose remaining from the neighboring node list of node in node listing
Neighbor node is added in recommended node list, until neighbor node quantity reaches the minimum forwarding number of degrees in recommended node list
Value；
3) when not meeting the recommended node of search request in the local interest concordance list and knowledge index table of node,
Expand search range, increases the numerical value of minimum forwarding degree.Random selection meets newest minimum forwarding from the neighbor node of node
Desired neighbor node is spent as recommended node, is added in recommended node list；
4) neighbor node forwarding inquiries message is selected from recommended node list successively, until list is empty for recommended node.
Fig. 3 is the embodiment that recommended node forwarding inquiries message is obtained by interest index and knowledge index.Destination node A
There are 7 connecting node B, C, D, E, F, G, H, wherein A and B, C, D have common interest vectorE and F has common interest vectorAssuming that minimum forwarding degree is 2, max-forwards degree is 3, index regulatory factor λ=
0.7.If it is " Google " to inquire subject key words, A obtains interdependent node B, C, D from interest concordance list, and by matching text
Gear number amount falls to sort, and calculates the practical forwarding degree k of three nodesB=3, kC=3, kD=2.Recommended node quantity is less than for 0 when beginning
kB=3, therefore B is selected, recommended node quantity is 1 no more than k at this timeC=3, C are selected.Recommendation number of nodes is 2 after selecting C
With kD=2 is equal, stops selection.Because recommended node quantity does not reach max-forwards degree 3, knowledge index table is then inquired,
There is no the node of matching inquiry keyword " Google " in knowledge index table, and recommended node quantity has reached minimum forwarding degree
2, stop selection course, final B, C are chosen as recommended node.If it is " Baidu " to inquire subject key words, A is indexed from interest
Recommended node C is obtained in table then to be pushed away from knowledge index table because recommended node quantity is not above max-forwards degree
Node E is recommended, recommended node quantity is 2 at this time, reaches minimum forwarding degree requirement, stops selection course, final C, E are chosen as recommending
Node.If inquiry subject key words be " Bing ", the node that interest concordance list and knowledge index table are not all met the requirements, then with
Machine selects G, H node as recommended node.If inquiry subject key words be " iPhone ", it not in the interest vector of A, then from
Interdependent node F and E, practical forwarding degree k are obtained in knowledge index tableF=3, kE=2, F and E are selected as recommended node, are turned
Hair number of nodes reaches minimum forwarding degree, and selection terminates.Query messages are forwarded to recommended node.
The technique effect of search method proposed by the present invention is illustrated below by experiment.
Experimental evaluation index：Average recall rate (Average Recall), forwarding expense (Number of
Transferring Query Message), recall rate (the Recall Per Transferring of each query messages
Query Message)。
Average recall rate (Average Recall)：Search in successful number of documents and network that line node is all to be met
The ratio of search request number of documents, averages, which is weighted average, and weight coefficient is defined as inquiring every time and be recalled
Rate accounts for all ratios for recalling rate score summation, and range is between zero and one.
Forward expense (Number of Transferring Query Message)：The forwarding number of query messages in network
The average value of amount, the value are weighted average.
The recall rate (Recall Per Transferring Query Message) of each query messages：Recall rate with
The ratio of expense is forwarded, which can be with the performance of evaluation technical proposal.
Simulating scenes are arranged：
1000 network nodes are generated, each node is bi-directionally connected with the holding of other 4 nodes at random, and each node has eight
Side is connected to other nodes, and initial network is unicom.Generate the vector for including 1024 subject key words.Generation 32 is emerging
Inclination amount, each interest include 32 keywords, and the keyword in interest vector is randomly selected from 1024 subject key words.
A node being assigned in network is randomly selected from 32 interest vectors, repeats the process, until each in network
Node is owned by an interest vector.3000 documents are generated, each document includes 8 keywords, and document keyword is from 1024
It is randomly selected in a subject key words.By all documents storage to network node.It is 30 days to emulate number of days, emulates 2000 daily
It is secondary, emulation total degree 60000 times.Daily (2000 times) for experimental evaluation index calculate average value.Emulation random selection every time
One network node generates query messages, and query messages hop count TTL is 3.
Fig. 4, Fig. 5, Fig. 6 are simulation result.As seen from Figure 4, technical solution IESLP proposed by the present invention is averagely recalled
Rate is higher than ESLP and NeuroGrid.As seen from Figure 5, the forwarding expense ratio ESLP of IESLP of the invention is low, than
NeuroGrid high.As seen from Figure 6, the recall rate that each query messages of IESLP of the invention obtain be higher than ESLP and
NeuroGrid。
In conclusion IESLP technical solutions performance proposed by the present invention is better than ESLP and NeuroGrid.
The above is used only for description technical scheme of the present invention and specific embodiment, is not intended to limit the present invention
Protection domain, it should be understood that under the premise of without prejudice to substantive content of the present invention and spirit, those skilled in the art make any repair
Change, improve or equivalent replacement etc. is fallen in protection scope of the present invention.
Claims (7)
1. a kind of reciprocity social networks document retrieval method of knowledge based study, which is characterized in that include the following steps：
Step A, node establish interest index and knowledge index, including：During file retrieval, node is from the identical mesh of interest
It marks node and obtains the similar knowledge store of interest in local interest concordance list, while obtaining according to searching keyword and being closed with inquiry
The matched knowledge store of keyword is in Indigenous knowledge concordance list；
Step B, node obtain neighbor node by local interest index and knowledge index and are used as recommended node, and to recommended node
Forwarding inquiries message, including：When node needs the query messages for forwarding other nodes to send over, local interest concordance list is inquired
The neighboring node list for including matching inquiry keyword document is obtained with knowledge index table, according to matching number of documents calculations list
The correlation coefficient of middle neighbor node, and minimum forwarding degree and max-forwards degree is combined to calculate adaptively forwarding degree, then foundation
The adaptive forwarding degree selection recommended node of neighbor node carries out query messages forwarding in list；
It is using neighbor node as the determination method of recommended node in the step B：If the adaptive of neighbor node turns in list
Hair number of degrees value is more than the quantity for having selected recommended node, then the neighbor node is chosen as recommended node；
The process that the step B selections recommended node carries out query messages forwarding is as follows：
Step 1) is then selected according to searching keyword from interest concordance list in the interest vector that searching keyword is included in node
It selects node to be added in recommended node list, when recommendation list interior joint lazy weight max-forwards number of degrees value, then according to inquiry
Keyword selects node to be added in recommended node list from knowledge index table, when recommendation list interior joint lazy weight is minimum
Number of degrees value is forwarded, then is added in recommended node list from the remaining node of random selection in neighbor list；
Step 2) is not when inquiry subject key words are in the interest vector of present node, then according to searching keyword from knowledge index
In table select node be added in recommended node list, when recommendation list interior joint lazy weight minimum forward number of degrees value, then from
Remaining node is randomly choosed in neighbor list to be added in recommended node list；
Step 3) expands search model when not meeting the recommended node of search request in interest concordance list and knowledge index table
It encloses, increases the numerical value of minimum forwarding degree；Random selection meets newest minimum forwarding degree from the connection neighbor node of present node
It is required that node as recommended node, be added in recommended node list；
Step 4) selects node forwarding inquiries message from recommended node list successively, until list is empty for recommended node.
2. a kind of reciprocity social networks document retrieval method of knowledge based study according to claim 1, which is characterized in that institute
State step A establish interest index and knowledge index process it is as follows：
Step 1), when query node is identical with destination node interest, interest vector of the query node according to destination node feedback
Keyword and destination node include that the number of documents of interest vector keyword establishs or updates local interest concordance list；
Step 2), when query node and destination node interest difference or query node with destination node interest identical and mesh
When marking node interest vector lists of keywords not comprising searching keyword, query node is crucial according to the inquiry of destination node feedback
Word and destination node include that the number of documents of searching keyword establishs or updates Indigenous knowledge concordance list；
Step 3), destination node is according to searching keyword statistical match number of documents.
3. a kind of reciprocity social networks document retrieval method of knowledge based study according to claim 1, which is characterized in that institute
It states and calculates correlation coefficient in step B, specially：
The correlation coefficient of i-th of the node obtained from interest concordance list or knowledge index table according to searching keyword isWherein i=1,2 ..., n, n are to be obtained from interest concordance list or knowledge index table according to searching keyword
Neighbor node number, the molecule m of item on the right of equationiFor i-th of node and the matched number of documents of searching keyword, denominatorFor the matching number of documents of node that is obtained from interest concordance list or knowledge index table according to searching keyword and.
4. a kind of reciprocity social networks document retrieval method of knowledge based study according to claim 1, which is characterized in that institute
It states and calculates the value of adaptive forwarding degree in step B and be：
The adaptive forwarding degree k of i-th of nodei=Round (ri λ(dmax-dmin))+dmin, wherein i=1,2 ..., n, riIt is i-th
The correlation coefficient of a node, dminFor minimum forwarding degree, i.e., at least by the neighbor node number of selection forwarding inquiries message, dmax
For max-forwards degree, i.e. the maximum node number by selection forwarding inquiries message, λ is index regulatory factor, range 0~1 it
Between, Round functions are bracket function.
5. a kind of reciprocity social networks document retrieval method of knowledge based study according to claim 4, which is characterized in that institute
State minimum forwarding degree dmin=2, max-forwards degree dmax=3, index regulatory factor λ=0.7.
6. a kind of reciprocity social networks document retrieval method of knowledge based study according to claim 1, which is characterized in that institute
It is the identical Hash table structure of structure to state interest concordance list and the knowledge index table, and keyword constitutes the key of Hash table, packet
The neighboring node list of the document of keyword containing matching inquiry constitutes the cryptographic Hash of key.
7. a kind of reciprocity social networks document retrieval method of knowledge based study according to claim 6, which is characterized in that institute
It includes two domains to state the element in list, and a domain stores information of neighbor nodes, and it includes pair that another domain, which stores the neighbor node,
The number of documents for the Keywords matching answered；The element comprising information of neighbor nodes presses number of files magnitude from big to small in the list
Bit-reversed, randomly selected node acquiescence matching number of files magnitude is 0, comes list end.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510955213.4A CN105447188B (en) | 2015-12-17 | 2015-12-17 | A kind of reciprocity social networks document retrieval method of knowledge based study |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510955213.4A CN105447188B (en) | 2015-12-17 | 2015-12-17 | A kind of reciprocity social networks document retrieval method of knowledge based study |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105447188A CN105447188A (en) | 2016-03-30 |
| CN105447188B true CN105447188B (en) | 2018-11-06 |
Family
ID=55557364
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510955213.4A Active CN105447188B (en) | 2015-12-17 | 2015-12-17 | A kind of reciprocity social networks document retrieval method of knowledge based study |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105447188B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106534481A (en) * | 2016-09-28 | 2017-03-22 | 努比亚技术有限公司 | Image or video sharing system and method |
| CN109658277B (en) * | 2018-11-30 | 2022-12-27 | 华南师范大学 | Academic social network friend recommendation method, system and storage medium |
| CN111694919B (en) * | 2020-06-12 | 2023-07-25 | 北京百度网讯科技有限公司 | Method, device, electronic equipment and computer readable storage medium for generating information |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101232415A (en) * | 2007-01-22 | 2008-07-30 | 华为技术有限公司 | Equity network node visit apparatus, method and system |
| CN102929914A (en) * | 2012-09-19 | 2013-02-13 | 浙江大学 | Mobile map service searching method based on P2P (point to point) node scheduling |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080052284A1 (en) * | 2006-08-05 | 2008-02-28 | Terry Stokes | System and Method for the Capture and Archival of Electronic Communications |
-
2015
- 2015-12-17 CN CN201510955213.4A patent/CN105447188B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101232415A (en) * | 2007-01-22 | 2008-07-30 | 华为技术有限公司 | Equity network node visit apparatus, method and system |
| CN102929914A (en) * | 2012-09-19 | 2013-02-13 | 浙江大学 | Mobile map service searching method based on P2P (point to point) node scheduling |
Non-Patent Citations (2)
| Title |
|---|
| Homingspread: community home-based Multi-copy Routing in Mobile Social Networks;Mingjun Xiao等;《IEEE》;20141231;第1-9页 * |
| 基于节点兴趣的P2P信息搜索机制研究与实现;宫月;《中国优秀硕士学位论文全文数据库 信息科技辑》;20080115;第I138-470页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105447188A (en) | 2016-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Leveraging social networks for P2P content-based file sharing in disconnected MANETs | |
| Upadrashta et al. | Social networks in peer-to-peer systems | |
| CN101378409A (en) | Optimal operation of hierarchical peer-to-peer networks | |
| CN105447188B (en) | A kind of reciprocity social networks document retrieval method of knowledge based study | |
| Xu et al. | A path-traceable query routing mechanism for search in unstructured peer-to-peer networks | |
| Ciobanu et al. | Data dissemination in opportunistic networks | |
| Saravanan et al. | An effective model for QoS assessment in data caching in MANET environments | |
| CN102377826A (en) | Method for optimal placement of unpopular resource indexes in peer-to-peer network | |
| Fraigniaud et al. | Combining the use of clustering and scale-free nature of user exchanges into a simple and efficient p2p system | |
| Liu et al. | Efficient resource discovery in self‐organized unstructured peer‐to‐peer networks | |
| Yeferny et al. | An efficient peer-to-peer semantic overlay network for learning query routing | |
| CN113934933A (en) | Low-delay message forwarding method, equipment and storage medium | |
| Hautakorpi et al. | A feasibility study of an arbitrary search in structured peer-to-peer networks | |
| Xu et al. | A search strategy for social resource in decentralized social networks | |
| Tran et al. | Multi-keyword based information routing in peer-to-peer networks | |
| Zhang et al. | A social network based file sharing system in mobile peer-to-peer networks | |
| Khatibi et al. | Dynamic multilevel feedback-based searching strategy in unstructured peer-to-peer systems | |
| Gu et al. | ASAP: An advertisement-based search algorithm for unstructured peer-to-peer systems | |
| Nguyen et al. | A socially inspired peer-to-peer resource discovery service for delay tolerant networks | |
| Nedos et al. | Probabilistic discovery of semantically diverse content in manets | |
| Nakashima et al. | Scalable tree-based consistency maintenance in heterogeneous P2P file sharing systems | |
| Wan et al. | A hybrid P2P overlay network for non-strictly hierarchically categorized content | |
| Chen et al. | An adaptive expanding antbudget search algorithm for unstructured P2P networks | |
| CN105808788B (en) | The data search method of big data system | |
| Gao et al. | A collaborative self-governing privacy-preserving wireless sensor network architecture based on location optimization for dynamic service discovery in MANET environment |
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 |