Abstraction: Question and Answer system is more popular in today’s universe. Increasing demand of speedy response to non factual questions like advice, recommendations are non solved absolutely by web hunt engines like Google. So, proposed Question and Answer system fulfill demands of non factual questions by using information on societal web. Existing system in market are depend on whatever informations already available in database, so new questions non acquire satisfied reply from such system. By using user’s societal information, Question and Answer system give satisfied reply to user. System enables user to send on inquiry to possible respondents in their friend list in decentralised mode. First order logic representation technique is used for ciphering involvement ID of user in different Fieldss. The inquiry is forwarded to merely that friend who able to give best reply based on similar involvement therefore cut downing calculation cost of nomadic nodes. By utilizing Question and Answer system based on societal web user can acquire high quality replies.
Index Footings– Question Answer System, Social based answering, reply form societal sites, non factual inquiries
The Internet is an of import beginning of information, and the sum of informations on the Internet is huge and invariably turning. Users rely on hunt engines to happen specific information within this cognition base. Search engines such as Google and Bing do a goodJOb of indexing web pages and supplying users with pages relevant to their hunt questions. However, there are some specific inquiries that are non suited for hunt engines. For illustration, aˆ•Where is the best eating house to acquire veg nutrient in Pune part? Q & A ; A systems have been developed to turn to this peculiar category of non-factual inquiries. Q & A ; A systems besides preserve all old inquiries and replies, therefore moving as a depository for information retrieval.
Question & A ; Answer system is featured by three advantages [ 1 ] :
( 1 )Decentralized-Rather than trusting on a centralized waiter, each node identifies the possible respondents from its friends, therefore avoiding the question congestion and high waiter bandwidth and care cost job.
( 2 )Low cost-Rather than airing a inquiry to all of its friends, an inquirer identifies the possible respondents.
( 3 )Quick response– An inquirer identifies possible respondents from his/her friends based on their past reply quality and replying action to his/her inquiries.
Online societal webs, unlike the Web, are organized around users. ToJoin an on-line societal web, users create a profile, publish content and construct links to anyone they want to tie in with. As a consequence, online societal webs have become the sites for keeping societal relationships, for happening users with similar involvements and for turn uping content and cognition that has been contributed or endorsed by other users.The great advantages that online societal webs bring to users include:
( 1 ) Low cost or no cost to entree the contained resources, with the Web-based nature of these webs intending informations is publically available ;
( 2 ) The sharing of cognition or experiences ;
( 3 ) Provision of up-to-date information ;
( 4 ) Ubiquity. Because of their value and advantages, the survey of these societal web web sites is of great importance to modern society.
Question Answering ( QA ) portal such as Yokel! Answers ( hypertext transfer protocol: //answers.yahoo.com ) , is an illustration of on-line collaborative societal webs. The chief intent of a collaborative societal web is to portion the cognition that users possess.
To heighten the inquirer satisfaction on the Q & A ; A sites, late, emerging research attempts have been focused on societal web based Q & A ; A systems, in which users station and reply inquiries through societal web maintained in a centralised waiter. As the respondents in the societal web know the backgrounds and penchant of the inquirers, they are willing and able to supply more trim and individualized replies to the inquirers. The social-based Q & A ; A systems can be classified into two classs: broadcasting-based and centralized. The broadcasting-based systems broadcast the inquiries of a user to all of the user’s friends. In the centralised systems since the centralized waiter concepts and maintains the societal web of each user, it searches the possible respondents for a given inquiry from the asker’s friends, friends of friends and so on.
However, the old broadcast medium and centralized methods are non suited to the nomadic environment, where each nomadic node has limited resources. Airing inquiries to a big figure of friends can non vouch the quality of the replies.
A. Question Processing
Social Network based Question Answer system ( SNQA ) incorporates an on-line societal web, where nodes connect each other by their societal links. A enrollment waiter is responsible for user enrollment. Each user has an involvement
ID, which represents his/her involvement. Users who have been willing to reply inquiries and provided high quality replies to nodei’s inquiries antecedently are more likely to be willing to reply nodeI’s inquiries and supply high quality replies. Therefore, SNQA has a metric best respondent( BA( chi,J))that measures likeliness of nodeJto be able and willing to reply nodeI’s inquirychiwith a high quality reply. It is determined by the involvement similarity( S( chi,J))between the inquiryQI’s involvement and nodeJ’s involvement every bit good as the reply quality ( Q( I,J)) of nodeJto node i’s old inquiries.
B.Question/User Interest Function
When a user first uses the SNQA system, s ( he ) is required to finish his/her societal profile such as involvements, professional
background and so on. Based on the societal information, the
enrollment waiter recommends friends to the user, and the
user so adds friends into his/her friend list. Each user
locally shops his/her ain profile and involvement ID, and friend
list and their involvement IDs and answer quality values. Each
user calculates his/her ain involvement ID on his/her sociAl
information and sends it to their friends. To cipher involvement
ID, a node first drives the first-order logic representation
( FOL ) from its societal information, so conducts first-order
logic illation to deduce its involvements, from which it decides
the involvement ID.
To parse a inquiry, the node first processes the inquiry
utilizing natural linguistic communication processing ( NLP ) , and so represents the inquiry in the FOL format and uses the FOL
illation to deduce the question’s involvements. Finally, it
transforms the inquiry to a inquiry ID in the signifier of a
numerical twine. After a node I parses its initiated inquiry
chito a inquiry ID, it calculates the involvement similarity S (chi,J)
for each of its friendsJ?FI, where FIdenotes the set of
node i’s friends. It so calculates the best respondent value
( BA(chi,J)) for each friendJby uniting S(chi,J)and reply
quality from friendJ( Q( I,J)) .Finally, node I choose top K
friends that have the highest Barium(chi,J)values to direct the
inquiry. By comparing the similarity between a question’s
ID and its friend’s involvement ID, a node can place its friends
that are able to reply inquiries.
C.First-order Logic Inference
The FOL [ 12 ] illation constituent consists of three parts:
( 1 ) fuzzy database, ( 2 ) regulations and maxims, ( 3 ) illation engine. The end of the illation is to place node involvements represented by a numerical twine that can accurately stand for the capableness of a node to reply inquiries. The fuzzed database is used to hive away words that have relationships, including subset, assumed name ( x ) , related, with the information in profiles. For illustration, related ( film ) =movie, subset ( computing machine scientific discipline, algorithm ) , alias ( USA ) =US.The regulation and maxims provide basic expression for the inference.The illation engine checks the regulations and discoveries related but non obvious information. It sets each involvement as an illation end and physiques lattice illation construction, as shown in Figure 1, to link All the FOL symbols with the ends. Each node in the lattice is an FOL sentence structure symbol and the pointers represent the conjunction symbols that connect the symbols.
III. SIMILARITY VALUE CALCULATION
After user’s societal information and inquiries are transformed into numerical strings, the similarity between a user and a inquiry can be calculated based on two parts: Interest similarity between the user and inquiry, and answer quality between the inquiry transmitter and receiving system.
To measure the involvement similarity of a inquiry of user I (chi) and a userJ, we use a method proposed in [ 25 ] . We use IDchiand ????DJto denote the involvement strings of inquirychiand userJ, severally. We use n(chi,J)to denote the figure of involvements owned byIdahochibut non byIdahoJ; usage cubic decimeter(chi,J)to denote the figure of classs of involvement elements owned both byIdahochiandIdahoJ, and m(chi,J)the figure of classs of involvement elements owned byIdahoJbut non byIdahochi. Then the involvement similarity of inquirychiand userJis defined as:
Second(chi,J)=(. ( 1 )
The value of ( ??‘z??‘- , ??‘- ) ranges in the classical spectrum [ 0, 1 ] , and it represents the degree of likeliness that two strings under comparing are really similar. If two strings have complete imbrication ( n=m=0 ) , ??‘† ( ??‘z??‘- , ??‘- ) approaches 1 as the figure of common characteristics grows. The implicit in thought of Equation ( 1 ) is that two strings with longer complete imbrication should hold higher similarity than the two strings with less complete imbrication. In the instance of no imbrication (l=0) , the map approaches to 0 every bit long as the figure of non-shared entries grows. It indicates that two strings with a larger figure of entries and portion no common entries are more likely to hold smaller similarity than the two strings with a smaller figure of entries and portion no common entries.
B. Answer quality computation
Social intimacy value computation mechanisms are based on the whole societal web topology, which are energy devouring. It is even worse when the societal web dynamically alterations. Therefore, the topology base societal intimacy computation methods are non suited for energy stringent nomadic devices in SNQA. Performance of the SNQA mostly depends on the action and the cognition base of the users, userIconsiders the figure of standard replies from userJand their associated quality evaluations when ciphering the answer quality of userJ. we call it as feedback mechanism. For each received reply, an inquirer can rate the quality of the reply within evaluation graduated table R= [ 1,5 ] . The reply quality value is updated based on the figure of replies received from friendJduring each period T and the associated quality evaluation (R? [ 1, 5 ] ) . For thekthinquiry sent from nodeIto nodeJ, if nodeIreceives an reply from nodeJduringThymine, ??‘???‘?=1 ; otherwise, ??‘???‘?=0. The parametric quantity ??‘???‘? is used to stand for the willingness of nodeJto reply inquiries from nodeI. Then, the reply quality ( ??‘- , ??‘- ) is calculated by:
??‘„ ( ??‘- , ??‘- ) = ??›? . ( ??‘- , ??‘- ) + 1 ? ??›? . ??‘? ( ??‘???‘? . ??‘Y??‘? /??‘… ) ( ??‘???‘? = 0,1 ) . ( 2 )
Where ? ? [ 0, 1 ] is a muffling factor,rkis nodeI’s quality evaluation for thekthreply received from nodeJ. A larger ( ??‘- , ) implies that userJis willing and able to supply high-quality replies to userI. Sing the high dynamism of the societal webs, in
which the willingness of users to reply inquiries and the quality of replies from a user to another user may alter over clip, we add muffling factor ? into the reply quality computation.
Based on above subdivisions, for its generated or received inquirychithat it can non reply, nodeIcalculates the best reply metric of each of its friends. That is
???µ ( ??‘z??‘- , ??‘- ) = ??›???‘† ??‘z??‘- , ??‘- + ( 1 ? ??›? ) ??‘„ ( ??‘- , ??‘- ) ( 3 )
Where ??›? ? [ 0,1 ] is a parametric quantity used to set the weight of the similarity and answer quality. NodeIso selects the topKfriends that have the highest ( ??‘z??‘- , ??‘- ) values and forwards the inquiry to them. Social trust between two nodes decrease exponentially with distance. This relationship has been confirmed by other surveies [ 28, 29 ] . A decrease in societal distance between two individuals significantly increases the trust between them.
Algorithm 1 shows the pseudo codification of the procedure for the best respondent metric computation and best respondent choice conducted by nodeI. If nodeIdoes non have replies for its created inquiry during the clip matching to TTL, it resorts to the centralized waiter for the replies, where all users conduct Q & A ; A activities in online Q & A ; A sites.
Algorithm 1[ 11 ]
Pseudo codification of the best respondent designation executed by
1:Input signal:IdahoI,IdahoJ,Q(I,J) (J?Fi)
2:End product:top-Kbest respondents
3: //Sporadically update Q(I,J) (J?Fi)
4:foreach friendJin friend listFimake
5: UpdateQ(I,J) based on Equation ( 2 )
7:ifmake a inquiry or have a inquiry it can non
8:ifTTL& gt ;0so
9:foreach friendJin friend listFimake
10: CalculateSecond(chi,J) utilizingIdahochiandIdahoJbased on
Equation ( 1 )
11: CalculateBarium(I,J) utilizingQ(I,J) andSecond(chi,J) based on
Equation ( 3 )
12: Attention deficit disorderBarium(I,J) to a listList
14: QuickSort divider around theKthlargest component
15: Find the top-Kfriends holding the highestBarium(I,J)
17: Send the inquiry to theIdahoentifiedKfriends
20:ifdoes non have replies for its created inquiry
during the clip matching to TTLso
21: Resort to the centralized waiter for the replies
Line4-Line6 are used to sporadically update answer quality of each of its friends. Line8-Line13 calculates each friend’s best respondent metric and generates a list including all metric values. Line14-Line17 identifies the top-Kfriends with the highest best respondent metric values and direct inquiry to them. Answer quality ( ??‘- , ??‘- ) is pre-processed and lone involvement similarity ??‘† ??‘z??‘- , ??‘- demand to be calculated at tally clip. The ( ??‘z??‘- , ??‘- ) computation has a clip complexness of ??‘‚ ??????‘- . As the figure of keywords in a inquiry is by and large really little, the computation of ??‘† ( ??‘z??‘- , ??‘- ) should take a short clip and costs small computation resources of the nomadic devices. This top-Kfriend choice algorithm has a clip complexness of??‘‚ ( | FI| ) .
SNQA systems are used by a big group of people for intents such as information retrieval, academic aid, and treatment. The turning importance of SNQA systems has led to legion research developments that are directed toward doing SNQA systems more effectual. The motive for this study is to analyze design of SNQA system. Besides it describe algorithm that used for best reply choice. Finally, SNQA system can be viewed as alternate solution to seek engines.
[ 1 ] Google. hypertext transfer protocol: //www.google.com
[ 2 ] B. M. Evans and E. H. Chi. An detailed theoretical account of societal hunt Information Processing & A ; Management, 2009.
[ 3 ] E. Amitay, D. Carmel, N. Har’El, S. Ofek-Koifman, A. Soffer, S. Yogev, and N. Golbandi. Social hunt and find
utilizing a incorporate attack. InProc. of HT, 2009.
[ 4 ] X. Yan. Smallblue: Social web analysis for expertness hunt and corporate intelligence. InProc. of ICDE, 2009.
[ 5 ] Harper, D. Raban, S. Rafaeli, andJoule. Konstan. Forecasters of reply quality in online Q & A ; A sites. InProc. of SIGCHI, 2008.
[ 6 ] M. R. Morris,Joule. Teevan, and K. Panovich. What do people inquire their societal webs, and why? : a study survey of position message Q & A ; A behaviour. In Proc. of CHI, 2010.
[ 7 ] J. Teevan, M. R. Morris, and K. Panovich. Factors impacting response measure, quality, and velocity for inquiries asked via societal web position messages. InProc. of AAAI, 2011.
[ 8 ] M. Richardson and R. W. White. Supporting synchronal societal Q & A ; a throughout the inquiry lifecycle. InProc. ofWWW, 2011.
[ 9 ] J. Raacke andJoule. Bonds-Raacke. MySpace and Facebook: using the utilizations and satisfactions theory to researching friend-networking sites.CyberPsychology & A ; Behavior, 2008.
[ 10 ] M. R. Morris,Joule. Teevan, , and K. Panovich. A comparing of information seeking utilizing hunt engines and societal webs. InProc. of ICSWM,2010.
[ 11 ] Z. Li, H. Shen, G. Liu, andJoule. Li. SNQA: A Distributed Context-AwareQuestion Answering System Based on Social Networks. InProc. ofICDCS, 2012.
[ 12 ] M. Kirsten and S. Wrobel. Widening K-Means Clustering to First-Order Representations. InProc. of ICILP, 2000.
[ 13 ] Z. Li and H. Shen. Soap: A societal web aided personalized and effectual Spam filter to clean your E-mail box. In Proc. of INFOCOM, 2011.
[ 14 ] E-commerce. hypertext transfer protocol: //en.wikipedia.org/wiki/E-commerce.
[ 15 ] C. Lampe,Joule. Vitak, R. Gray, and N. B. Ellison. Percepts of facebook’s value as an information beginning. InProc. Of CHI, 2012
No related essays.