Best Keyword Cover Search Using Distance and Rating

Transcription

1 Best Keyword Cover Search Using Distance and Rating Mr. Vishal D. Kolekar, Prof. Ajay Kumar Gupta Department of Computer Engineering, SPs Institute Of Knowledge College Of Engineering, Savitribai Phule Pune University, Maharashtra. Abstract Spatial databases are stores the information about the spatial objects which are associated with the keywords to indicate the information such as its business/services/fea-tures. Very important problem known as closest keywords search is to query objects, called keyword cover. In closest keyword search, it covers a set of query keywords and minimum distance between objects. From last few years, keyword rating increases its availability and importance in object evaluation for the decision making. This is the main reason for developing this new algorithm called Best keyword cover which is considers inter-distance as well as the rating provided by the customers through the online business review sites. Closest keyword search algorithm combines the objects from different query keywords to generate candidate keyword covers. Baseline algorithm and keyword nearest neighbors expansion algorithms are used to find the best keyword cover. The performance of the closest keyword algorithm drops dramatically, when the number of query keyword increases. To solve this problem of the existing algorithm, this work proposes generic version called keyword nearest neighbor expansion which reduces the resulted candidate keyword covers. Index Terms Spatial database, point of interests, keywords, keyword rating, keyword cover. I. INTRODUCTION Now a days, use of mobile computing increases. Inspired by the mobile computing, the spatial keywords search problem has attracted much attention recently because of locationbased services and wide availability of extensive digital maps and satellite imagery. Spatial objects indicates the information such as its business/services/features which are associated to keyword(s). In spatial database, each tuple represents a spatial object. The main idea behind the spatial keywords search is to identify spatial object(s) which are associated with keywords relevant to a set of query keywords which are close to each other and/or close to the query location. This problem has unique value in vari-ous applications because users requirements are often expressed as multiple keywords. In existing, spatial keyword search problem have been studied because of the value of the special keyword search in practice. This paper investigates a generic version of mck query, called Best Keyword Cover (BKC) query, which considers interobjects distance as well as keyword rating. It is motivated by the observation of increasing availability and importance of keyword rating in decision making. Millions of businesses/services/features around the world have been rated by customers through online business review sites such as Yelp, City search, ZAGAT and Dianping, etc. For example, a restaurant is rated 65 out of 100 (ZAGAT.com) and a hotel is rated 3.9 out of 5 (hotels.com). According to a survey in 2013 conducted by Dimensional Research (dimensionalresearch.com), an overwhelming 90 percent of respondents claimed that buying decisions are influenced by online business review/rating. Due to the consideration of keyword rating, the solution of BKC query can be very different from that of mck query. This work develops two BKC query processing algorithms, baseline and keyword-nne. The baseline algorithm is inspired by the mck query processing methods Both the baseline algorithm and keyword-nne algorithm are supported by indexing the objects with an R*-tree like index, called KRRtree. In the baseline algorithm, the idea is to combine nodes in higher hierarchical levels of KRR*-trees to generate candidate keyword covers. Then, the most promising candidate is assessed in priority by combining their child nodes to generate new candidates. Even though BKC query can be effectively resolved, when the number of query keywords increases, the performance drops dramatically as a result of massive candidate keyword covers generated. To over-come this critical drawback, we developed much scalable keyword nearest neighbor expansion (keyword-nne) algorithm which applies a different strategy. Keyword-NNE selects one query keyword as principal query keyword. The objects associated with the principal query keyword are principal objects. For each principal object, the local best solution (known as local best keyword cover (lbkc)) is computed. Among them, the lbkc with the highest evaluation is the solution of BKC query. Given a principal object, its lbkc can be identified by simply retrieving a few nearby and highly rated objects in each nonprincipal query keyword (two-four objects in average as illustrated in experiments). Compared to the baseline algorithm, the number of candidate keyword covers generated in keyword- NNE algorithm is significantly reduced. The in-depth analysis re-veals that the number of candidate keyword covers further processed in keyword-nne algorithm is optimal, and each keyword candidate cover processing generates much less new candidate keyword covers than that in the baseline algorithm.

2 The goal is to rank the methods, so we only report here on the binary comparisons that allowed us to determine the ordering of the four methods (excluding redundant comparisons).our current goals are to allow explicit queries, and to rank document results with the objective of maximizing the coverage of all the in the spatial database, while minimizing redundancy in a short list of the best keyword search. A keyword cover of keyword that is the word related to that keyword, and cover keyword is called to be the best keyword for the search find valuable search and ranking, without interrupting the conversation flow, thus ensuring the usability of our system. In the future, this will be tested with human users of the system within real - life meetings. A. Motivation Driven by mobile computing, location-based services and wide availability of extensive digital maps and satellite imagery (e.g., Google Maps and Microsoft Virtual Earth services), the spatial keywords search problem has attracted much attention recently. A tourist who plans to visit a city may have particular shopping, dining and accommodation needs. It is desirable that all these needs can be satisfied without long distance travelling. User want spatial objects which are associated with the query keywords. Millions of businesses/services/features around the world have been rated by customers through online business review sites. While retrieving the query objects that are closed to each other and/or close to the query location. B. Problem Statement When the number of query keyword increases, the performance drops dramatically as a result of massive candidate keyword cover generated. To overcome this critical problem, proposed system developed much scalable keyword nearest neighbour expansion (keyword-nne) algorithm which applies a different strategy. II. REVIEW OF LITERATURE A. Ruicheng Zhong, JuFan, Guoliang Li, Kian-Lee Tan, LizhuZhou, Location-Aware Instant Search. Now a days, mobile users widely uses the Location-Based Services (LBS). Location-based systems can work efficient if user enters the complete keyword, otherwise it was shown incomplete output. It difficult to enter the complete keyword on mobile devices for get the correct relative result. To avoid this problem, proposed system studied about the locationaware search. It can be returned search answers as the user enters in queries letter by letter. In this paper, the main challenge is to provide the relevant answers speedily. Author uses a new index structure, prefix-region tree (called PR-tree), that can be support to provide the speedily results to the users. PR-Tree is a tree-based index structure which seamlessly integrates the textual description and spatial information to index the spatial data. Using the PRTree, authors develop efficient algorithms to support single prefix queries and multikeyword queries[6]. B. Xin Caoy, Gao Congy, Christian S. Jensenz, Retrieving Top-k Prestige Based Relevant Spatial Web Objects. The location-aware keyword query returns ranked objects that are near a query location and that have textual descriptions that match query keywords. There are many mobile applications and traditional services uses this type of query, e.g. Yellow pages and Maps services. In previous work, ranked query returns independent potential results. Ranking is very important in decision making. However, a relevant result object with nearby objects that are also relevant to the query is likely to be preferable over a relevant object without relevant nearby objects. The paper proposes the concept of prestige-based relevance to capture both the textual relevance of an object to a query and the effects of nearby objects. Based on this, a new type of query, the Location-aware top-k Prestige-based Text retrieval (LkPT) query, is proposed that retrieves the topk spatial web objects ranked ac-cording to both prestige-based relevance and location proximity. We propose two algorithms that compute LkPT queries. Empirical studies with real-world spatial data demonstrate that LkPT queries are more effective in retrieving web objects than a previous approach that does not consider the effects of nearby objects; and they show that the proposed algorithms are scalable and outperform a baseline approach significantly[2]. C. Ian De Felipe, Vagelis Hristidis, Naphtali Rishe, Keyword Search on Spatial Databases. There are lots of applications that finds the objects nearest to the specified location which contains a set of keywords. Yellow pages required address and a set of keywords to get the results. Yellow pages returns a list of business/features/services whose description contains entered keywords, ordered by their inter-object distance from the specified location. In this paper, author studied problems of nearest neighbor search on location data and keyword search on text data separately. There is no any method that returns answer to the best for spatial and keyword queries which is related to the same. In this paper, author proposed an efficient algorithm that returns top-k spatial keyword queries. Proposed system introduces indexing structure called Information retrieval R-Tree which is combination of R-tree with superimposed text signatures. Algorithm returns the answer from IR2-tree which construct and maintain by the algorithm to the keyword queries. Proposed algorithms are superior performance and excellent scalability to the previous work experimentally[5]. D. Zhisheng LI, Ken C.K. LEE, Baihua ZHENG, Wang- Chien LEE, Dik Lun LEE, IR-tree: An Efficient Index for Geographic Document Search Geographic search engine returns documents which are very close textually and spatially to the query keywords. Retrieved documents are ranked according to their joint textual and spatial relevance to the entered query. Existing indexing scheme inefficient in answering spatial queries because of lacking in index simultaneously handle both the textual and location aspect. In this proposed system, author proposes new

3 index called IR-tree that combines with a top-k document facilitates four major tasks in document searches, textual filtering, spatial filter-ing, relevance computation, and document ranking. These four tasks are used in this algorithm in a fully integrated manner. Also, this algorithm adopt different ranks on textual and spatial relevance of documents at the run time[7]. Roy and Chakrabarti describes user s often search spatial database like yellow page data using keywords to businesses near their current location [8, 9]. Such searches are increasingly performed from mobile devices. Typing the entire query is cumbersome and prone to errors, especially from mobile phones. We address this problem by introducing type-ahead search functionality on spatial databases. Like keyword search on spatial data, type-ahead search needs to be location-aware, i.e., with every letter being typed, it needs to return spatial objects whose names (or descriptions) are valid completions of the query string typed so far and which rank highest in terms of proximity to the user s location and other static scores. Existing solutions for type-ahead search cannot be used directly as they are not location-aware. We show that a straightforward combination of existing techniques for performing type-ahead search with those for performing proximity search perform poorly [12]. We propose a formal model for query processing cost and develop novel techniques that optimize that cost. Our empirical evaluations on real and synthetic datasets demonstrate the effectiveness of our techniques. To the best of our knowledge, this is the rest work on locationaware type-ahead search. III. EXISTING SYSTEM Some existing works focus on retrieving individual objects by specifying a query consisting of a query location and a set of query keywords (or known as document in some context). Each retrieved object is associated with keywords relevant to the query keywords and is close to the query location. The approaches proposed by Cong et al. and Li et al. employ a hybrid index that augments nodes in non-leaf nodes of an R/R*-tree with inverted indexes. In virtual br*-tree based method, an R*-tree is used to index locations of objects and an inverted index is used to label the leaf nodes in the R*-tree associated with each keyword. Since only leaf nodes have keyword information the mck query is processed by browsing index bottom-up. A. Limitations of Existing System When the number of query keywords increases, the performance drops dramatically as a result of massive candidate keyword covers generated. The inverted index at each node refers to a pseudodocument that represents the keywords under the node. Therefore, in order to verify if a node is relevant to a set of query keywords, the inverted index is accessed at each node to evaluate the matching between the query keywords and the pseudo-document associated with the node. IV. SYSTEM OVERVIEW Some existing works focus on retrieving individual objects by specifying a query consisting of a query location and a set of query keywords (or known as document in some context). Each retrieved object is associated with keywords relevant to the query keywords and is close to the query location. The approaches proposed by Cong et al. and Li et al. employ a hybrid index that augments nodes in non-leaf nodes of an R/R*-tree with inverted indexes. In virtual br*-tree based method, an R*-tree is used to index locations of objects and an inverted index is used to label the leaf nodes in the R*- tree associated with each keyword. Since only leaf nodes have keyword information the mck query is processed by browsing index bottom-up. The spatial keyword search has received considerable attention from research community. Some existing works focus on retrieving individual objects by specifying a query consisting of a query location and a set of query keywords (or known as document in some context). Each retrieved object is associated with keywords relevant to the query keywords and is close to the query location. The similarity between documents are applied to measure the relevance between two sets of keywords. Since it is likely no individual object is associated with all query keywords, some other works aim to retrieve multiple objects which together cover all query keywords. While potentially a large number of object combinations satisfy this requirement, the research problem is that the retrieved objects must have desirable spatial relationship.authors put forward the problem to retrieve objects which 1) cover all query keywords, 2) have minimum inter-objects distance and 3) are close to a query location. The work study a similar problem called m Closet Keywords (mck). mck aims to find objects which cover all query keywords and have the minimum inter-objects distance. Since no query location is asked in mck, the search space in mck is not constrained by the query location. The problem studied in this paper is a generic version of mck query by also considering keyword rating of objects. A. Keyword-NNE The goal is to rank the methods, so we only report here on the binary comparisons that allowed us to Given a spatial database, each object may be associated with one or multiple keywords. Without loss of generality, the object with multiple keywords are transformed to mul-tiple objects located at the same location, each with a distinct single keyword. When further processing a candidate keyword cover, keyword- NNE algorithm typically gen-erates much less new candidate keyword covers compared to BF-baseline algorithm. Since the number of candidate keyword covers further processed in keyword-nne al-gorithm is optimal the number of keyword covers generated in BF-baseline algorithm is much more than that in keyword- NNE algorithm. In turn, we conclude that the

4 number of keyword covers generated in baseline algorithm is much more than that in keyword-nne algorithm. This conclusion is independent of the principal query keyword since the analysis does not apply any constraint on the selection strategy of principal query keyword. The keyword nearest neighbour expansion algorithm reduces the number of can-didate keyword covers generated. In-depth analysis reveals that the number of candidate keyword covers further processed in keyword-nne algorithm is optimal. The best keyword cover uses the distance between interobjects as well as the ratings provided by the customers. B. Algorithm 1) The Friend Matchiing Graph: Input: The Friend Matchiing Graph G. Output: Impact ranking vector r for all users. 1) for i= i to n do 2) r(0)n = n 1 3) end for 4) = 1 5) = e 9 6) while > do 7) for i= 1 to n do 8). rk+ j p 1 n rk(j)+ p w(i.j):rk(j) jw(i.j) p 9) If no data point was reassigned then stop, otherwise repeat from step 3). Advantages of K-Means Algorithm Fast, robust and easier to understand. Relatively efficient: O(tknd), where n is objects, k is clusters, d is dimension of each object, and t is iterations. Normally,k, t, d << n. Gives best result when data set are distinct or well separated from each other. 3) Mathematical Model: The goal is to rank the methods, so we only report here on the binary comparisons that allowed us to Set Theory S is the System, User Module U={U0, U1, U2} U0=Add Location U1= Query Keywords U2= Rating Object and Feedback Admin Module A={A0,A1,A2} A0=Add User A1=Add Description A2=Add user Location Clustering Module C={C0,C1,C2} C0=Clustering keywords C1=Search nearest result C2=Best Keyword Covered. 9) end for 10) = P n i=1 jrk + 1rkj 11) end while 12) return r. 2) K-means: Let X = {x1,x2,x3,..,xn } be the set of data points and V = {v1,v2,.,vc} be the set of centers. 1) Randomly select c cluster centers. 2) Calculate the distance between each data point and cluster centers. 3) Assign the data point to the cluster center whose distance from the cluster center is minimum of all the cluster centers. 4) Recalculate the new cluster center using: 5) V i = (1)P C i X i 6) C i j = 1 7) where, C i represents the number of data points in ith cluster. 8) Recalculate the distance between each data point and new obtained cluster centres. Fig. 1. Venn Diagram V. SYSTEM DESCRIPTION In proposed system, best keyword cover (BKC) query which is generic version of the mck (m Closest Keywords) query, which considers the distance between inter-objects as well as keyword rating provided by the customers. In proposed system, baseline algorithm and keyword nearest neighbor expansion algorithms are used. It is motivated by the observation

5 VII. SYSTEM RESULT Fig. 2. Block Diagram of Proposed System of increasing availability and importance of keyword rating in decision making. To overcome drawback of existing system, proposed system developed much scalable keyword nearest neighbor expansion (keyword- NNE) algorithm which applies a different strategy. We will use Spatial data module and User module where Spatial data contain the geographical data having latitude and longitude and User module data contain point of interest data which might be different for individual users. So we can say that Spatial module contain static data and User module contain dynamic data which ultimately brings dynamic data to user. Fig. 3. Upload Location Information A. Hardware Requirement Processor: Pentium-IV Speed: 1.1 GHz Ram: 1GB(min) Hard Disk: 80GB Monitor: SVGA. B. software Requirement VI. SYSTEM REQUIREMENT Operating System: windows Xp and Higher Software: Andriod studio Database: MySQL 5.5 Programming Language: PHP,Java Fig. 4. Keyword Search

6 Fig. 5. Keyword Covered Places Fig. 7. Graphical analysis The above graph shows that the efficiency of the proposed system. On X axis it shows system and on Y axis its shows number of related query result. The graph shows that the output of proposed system is more efficient as it contains more related query result as compare to mkc. IX. CONCLUSION Comparing the most of the relevant mck query,bkc query provides an advanced dimension to support more effective decision making. The introduced baseline algorithm is inspired by the methods for processing mckquery. K-maens clustering algorithms are also used. The baseline algorithm generates a large candidate keyword covers which leads to dramatic performance drop when more query keywords are given. The proposed keyword-(nne)algorithm applies the various processing schema i.e, finding local good solution for every object in a certain query keyword. As a consequence, the No. of candidate keyword covers generate significantly reduced. The analysis reveals that the number of candidate keyword covers which need to be further processed in keyword-(nne)algorithm is optimal and processing each keyword candidate cover typically generates less new candidate keyword covers in keyword- (NNE) algorithm than in the baseline algorithm. Fig. 6. Rating by user VIII. SYSTEM ANALYSIS X. FUTURE SCOPE The proposed system provides information related to the keyword entered by user according to inter object distance and ratings. In future we will make this search personalized by introducing user profiles with his previous search results and domain knowledge.

7 REFERENCES [1] Ke Deng, Xin Li, Jiaheng Lu et al. Best keyword cover search. IEEE Transactions on Knowledge and Data Engineering. 2015; 27(1). [2] X. Cao, G. Cong, C. Jensen. Retrieving top-k prestige-based relevant spatial web objects. Proc. VLDB Endowment. 2010; 3(1/2): p. [3] X. Cao, G. Cong, C. Jensen et al. Collective spatial keyword querying. In Proc. ACM SIGMOD Int. Conf. Manage. Data. 2011; p. [4] G. Cong, C. Jensen, D. Wu. Efficient retrieval of the top-k most relevant spatial web objects. Proc. VLDB Endowment. 2009; 2(1): p. [5] I. D. Felipe, V. Hristidis, N. Rishe. Keyword search on spatial databases. In Proc. IEEE 24th Int. Conf. Data Eng. 2008; p. [6] R. Hariharan, B. Hore, C. Li et al. Processing spatial keyword (SK) queries in ge-ographic information retrieval (GIR) systems. In Proc. 19th Int. Conf. Sci. Statist. Database Manage. 2007; 1623p. [7] Z. Li, K. C. Lee, B. Zheng et al. IRTree: An efficient index for geographic docu-ment search. IEEE Trans. Knowl. Data Eng. 2010; 99(4): p. [8] J. Rocha-Junior, O. Gkorgkas, S. Jonassen et al. Efficient processing of top-k spa-tial keyword queries. In Proc.12th Int. Conf. Adv. Spatial Temporal Databases. 2011; p. [9] S. B. Roy, K. Chakrabarti. Location-aware type ahead search on spatial databases: Semantics and efficiency. In Proc. ACM SIGMOD Int. Conf. Manage. Data. 2011; p. [10] D. Zhang, Y. Chee, A. Mondal et al. Keyword search in spatial databases: Towards searching by document. In Proc. IEEE Int. Conf. Data Eng. 2009; p [11] N. Mamoulis and D. Papadias, Multiway spatial joins, ACM Trans. Database Syst., vol. 26, no. 4, pp , [12] D. Papadias, N. Mamoulis, and B. Delis, Algorithms for querying by spatial structure, in Proc. Int. Conf. Very Large Data Bases, 1998, pp [13] D. Papadias, N. Mamoulis, and Y. Theodoridis, Processing and optimization of multiway spatial joins using r-trees, in Proc. 18th ACM SIGMOD-SIGACTSIGART Symp. Principles Database Syst., 1999, pp [14] J. M. Ponte and W. B. Croft, A language modeling approach to information retrieval, in Proc. 21st Annu. Int. ACM SIGIR Conf. [15] J. Rocha-Junior, O. Gkorgkas, S. Jonassen, and K. Norva g, Efficient processing of top-k spatial keyword queries, in Proc. 12th Int. Conf. Adv. Spatial Temporal Databases, 2011, pp Collaborative computing, INFOCOM 2008, pp