G1803054653

IOSR Journal of Computer Engineering (IOSR-JCE)
e-ISSN: 2278-0661,p-ISSN: 2278-8727, Volume 18, Issue 3, Ver. V (May-Jun. 2016), PP 46-53
www.iosrjournals.org
DOI: 10.9790/0661-1803054653 www.iosrjournals.org 46 | Page
A Review on Semantic Approach using Nearest Neighbor Search
S. Haripriya1
, A. Brahmananda Reddy2
, A. Prashanth Rao3
1
(Software Engineering, VNR Vignana Jyothi Institute of Engineering & Technology, India)
2
(Computer Science & Engineering, VNR Vignana Jyothi Institute of Engineering & Technology, India)
3
(Information Technology, Anurag Group of Institutions, CVSR, India)
Abstract: Information retrieval (IR) of acquiring information sources relevant to an information sources from
a group or unit. Ontology in IR field is used to represent an official domain description in addition a semantic
layer is added to the IRS. The idea depicts in relating semantics on the concepts of ontology using query words.
Semantic approach, focus to search effectively by assuming the searchers contextual meaning and purpose the
way they occur in terms. It aims to improve the search for hospitals with its specialties relevant to disease given
through text analysis. Distance from search location can be viewed in map view. A technique called spatial
inverted index to locate multidimensional information and draws algorithms which will acknowledge nearest
neighbor queries. Nearest neighbor can locate the hospitals that are closest to a given address.
Keywords: Context, Information Retrieval, Ontology, Semantic Approach, Spatial Inverted Index
I. Introduction
Information retrieval activity starts when a user enters a query. To the given query it generates a
method called Semantic based approach. The user provides with a phrase or context that is planned to gather
or search information that imply an object regarding whichever the searcher is trying to search for. Firstly, it
aims to improve the search for hospitals with specialties relevant to disease through text analysis.
Secondly, search locations distance can be viewed in map view. Nearest neighbor (NN) either called as
parallel search or nearest point search is a problem of searching nearest points. Extracting Nearest neighbor
includes conditions on geometric effects of objects. This process can be done by using K nearest neighbor
queries (KNN) and Location based services (LBS). KNN solutions are based to show efficient short distances.
A well coordinated index (signature files) for distance calculation and query transformation over great distances
is used. But this search concentrates mainly on distance metric, but not on text description, i.e. on context of
spatial objects in generating the query.
Context retrieval has two important indexing approaches, signature files and inverted files. To manage
spatial text queries is to merge two types they are nearest neighbor queries and text extraction. The algorithm
places an inverted index for all words, and again builds R*-tree for each context. The outstanding system is to
frame an inverted index above R*-trees. IR-tree consists of both inverted files and R-tree to generate K best
results which are maintained in a ranking scheme and administer location-aware context extraction. IR2-tree
which combines signature files and R-tree to address top-k spatial contextual queries. IR2-tree possesses a
pitfall of signature file incorrect hits. To eliminate this disadvantage, establish an advanced access system called
spatial inverted index to handle with multidimensional data which enhance traditional inverted index.
II. Literature Review
The semantic search particularly helpful in applications where the user hunts down the model of
reasonable occurrences, rather than hunting down "subjective" web pages. That is, the question indicates one or
more idea more often than not by utilizing watchwords. These questions are portrayed as exploration hunts. As
it were, an occasion of a hub in the model is a hypermedia representation of every page in the application. As a
rule, it is additionally helpful for site pages when connected with rich metadata [14].
Cluster measure
To set up between two related idea cases in a connection and the level of comparability is measured.
The comparability measure utilized is fundamentally the same to the bunch capacity utilized as a part of,
acquired by practicing that capacity for ideas that identify with each other. The likeness between idea occasion
Cj and idea case Ck demonstrates the equation below.
W(Cj,Ck) = ∑ nijk /∑ nij ---- (1)
Specificity Measure
The second measure is like the reverse space recurrence measure [18] broadly utilized as a part of
Information Retrieval (inspite of the fact that in I.R. The log capacity is regularly utilized). This measure is

helpful when the client needs to give the semantics of specificity or separation to the connection as mentioned
in Table No: 1. The accompanying equation was utilized for the specificity measure:
W(Cj,Ck) = 1 / nk --------- (2)
Table No: 1 Ranking Results
Application TestType Number Instance Positive Evaluator
Website DI 1 20 100%
Website DI 2 10 100%
Partinori 1 20 90%
Partinori 2 10 100%
Semantics are being incorporated into the search engine of the major search companies. The aim to
elucidate doubts regarding the different approaches for ranking results in Semantic Search. An overview of each
approach to discuss in brief about them as well as try to give a succinct explanation of the working of the
approaches. Further the advantages and disadvantages have been stated wherever possible. The information
boom has further aggravated the situation of World Wide Web. Searching has become a complex task. In the
purview of overcoming this difficulty has become more important. Semantic Search offers the possible solution
to this problem and different approaches of semantics are described in Table No: 2 [15].
Table No: 2 Different approaches of semantics
Authors Approaches Focus Association Determination Architecture Input Effectiveness
Rocha et al. Hybrid
Spread
Activation
Entity
based
Ranking
Combination of
Clustering measure and
Specificity
Measure
Stand Alone Keyword
query
Semantically
effective
Anyanwu et
al.
SemRank Relations
hip based
Ranking
Top-K ordering algorithm
and Annotation Path
Expression
Depends on
the
architecture of
SSARK
system
Query and
the level of
result
required
search
Effective on
small set.
Wei et al. Rank Entity
based
Ranking
Link Analysis
Based
Meta Keyword
query
Very effective
when compared
to
PageRank
Lamberti et al. Relation
based Page
rank
Relation
between
keywords
&
concepts
Page relevance
and scoring using sub graph
and ontology graph
Graph Based Set of
keywords,
concepts
Effective as it
interprets hidden
concepts behind
keywords
A text-free alphabet to define the query expression of accountable by the organization which utilizes a
hybrid data extraction for domain data extraction. The organization utilizes domain query elements to aid
analysis of template-based specification. In calculation among the new semantic approach (Hakia),
crowdsourcing approach (DuckDuckGo) with the trendy research strategy as Google, the system achieve for
complex information needs is satisfactorily in retrieving relevant results. A search that arrive on concepts and
popular in prescribing drug utilizes social media by epidemiologists conducting alive web application is at
present applicable and in usage [16].
Semantic matching in the ontology field results in input query and data. The query and data field
extracts occurances from the hybrid technique that is on merging. Semantic matching is focused on queries and
information fields, to design the better match and to result the managerial process. Hybrid ontology that
correlates to a basic ontology on the semantic web is plentiful to extract the files [17].
T(k) = (f . p)+((m . (1/(1+e-x) )/2) ---------- (3)
A client set least backings and chooses about which rules have high backing. Once the standards are
chosen, all are dealt with as the same, regardless of how high or how low their backing. In Uniform, whereas in
Skew, Zip f distribution is followed by their locations are dispersed. In this system a new way of providing all
details of the tablet along with distance to the user. In contrast with the existing system this system shows the
efficiency. The system will be helpful for the implementation of the SNNS (Semantic Nearest Neighbor Search)
system is designed [8].
The conceptual models and information (e.g., orders, database blueprints, ontologies) may be
processed by using chart diagrams. This takes into account the announcement and the arrangement of a bland
(semantic) coordinating issue autonomously of particular applied or information models, as done in both
COMA and Cupid. A structure like tree, whereas XML schemas, and classifications. Seldom trees are actual-
world system, hence, many advanced methods, constructing the chart view as blueprint in a tree description.
The access has established semantic matching with two key concepts, specifically:

The Concept of a label
The label implies the collection of files (data instances) which analyzes and encodes under a label.
The Concept at a node
The node describes the set of files (data instances) which would analyze below a node, given as it is in
an accurate position in a tree and that it has a certain label [18].
The IR2-Tree and in which way it is managed in the existence of data renew are imported. A dynamic
incremental algorithm is implemented to answer spatial keyword queries which uses the IR–Tree. Analytically
evaluated technique, that demonstrate its exceptional performance. Plentiful applications lack a web research
tool that can adequately aid novel categories of spatial query with a unified context search. Present answers for
this questions rather cause restrictive area utilization as a choice can't give ongoing answers. The circumstance
is cured by building up an entrance technique called the spatial inverted index (SIindex). It additionally can
perform catchphrase enlarged nearest neighbor search all together of many milliseconds of time and not just that
the SI-file is reasonably space conservative. In addition, SI-index as established a mechanism of a traditional
inverted index, and is promptly in-comparable in an economical search tool which employs enormous
parallelism, involve its actual modern merits [2].
The KNN Spatial Keyword Query Process Is Shown As Follows:
The input to the query q point, the limit object BO, the k parameter and context word Kw. By the
server kNN results are returned is recovered by BO call, Result (). Min-store is H that search implies agreeing
its separations to inquiry q. To begin with, the calculation develops the limit cell (BC) of the primary article p1
and analysis if q drops in BC (p1). No more, NN & check generates falls flat where p1 is not main. Something
else, p1 is confirmed as the NN principal and adds to the set Visited. The consequent for circle repeats through
each items in L (BO from the kNNs) which process the accompanying techniques: 1) if the last neighbor of
confirmed article (L[i]) has not been gone but still, that is embedded the into min-load H & the set Visited & 2)
that thinks about a following item from the set outcome (L[i+1]) along the highest point H. In the event that are
indistinguishable, L[i+1] are checked in following NN. Something else, confirmation comes up short and the
system returns false [2].
The IR-tree is used to extract a collection of objects with spatial web. Those groups of objects contain
query’s and objects are near to the query address & the distances of similar objects are low. The two
illustrations of the group keyword query, to search the set of objects that covers query keywords, then the
addition of their distances to the query location is reduced. Another one is to search a collection of objects with
the query keywords are highest than the distance among an object in a set of objects and query and high
distance among two objects in a group of objects is minimum. The Greedy algorithm which uses IR-tree to
decrease the search space is an appropriate answer to the described problem is followed. But, some application
query contain a small number of keywords, for this exact algorithm is used and it uses the dynamic
programming [20]. Cong et al assumed a key based closest neighbor inquiries which are like yet varies in how
item's writings assume a part in deciding the consequence of the inquiry. In particular, going for an IR tree, the
methodology figures the importance of an item panda question q of the archives. Pertinence to the score of the
Euclidean separation in the middle of p and q is then coordinated to figure a general closeness of q to p. Some
items which are most noteworthy comparability are restored. Along these lines, an article may in any case be in
the question result, despite the fact that its report does not contain all the inquiry watchwords. The technique
here accompanies object messages that are used in a Boolean predicate assessing, i.e., any question watchword
are absent in the article's record, should not return. Either one of its approaches subsumes the another, no two
bode well in various applications. There is no "halfway fulfillment", similar to the method of reasoning behind
the methodology. In geographic web look, every website page is doled out a geographic locale that is related to
the site page's substance [7]. Based on the pair of actual data and synthetic always the dimensionality 2, consists
of integers from 0 to 16; 383 with all axis. The category consists of two synthetic data sets: Skew and Uniform,
in contrast of the appropriation information focuses, & on either it is a relationship among the spatial
conveyance and articles' content archives. In particular, every information of 1 million focuses in a set. The
areas are consistently appropriated though in Skew, in Uniform, they take after the Zipf dissemination.
For two sets of information, the dictionary consists of words 200, and every word shows up in the
content reports of 50k focuses are described in Table No: 3. The distinction is the relationship with focuses of
words is totally irregular in Skew, while in Uniform, there is an example of ‘word-region’ focuses which are
spatially near have practically indistinguishable content reports [7].
Table No: 3 Dataset Statistics
Number of points Dictionary size Average no of objects for each word Average no of words for each object
Uniform 1000000 200 50000 10
Skew 1000000 200 50000 10
Census 20847 292255 33 461

The reverse nearest neighbor queries together with notion of keyword search. An IUR (Intersection
Union RTree) is a combinational index tree that answers the Reverse Spatial Textual k Nearest Neighbor
(RSTkNN) query that effectively associates location closeness with text is proposed. A branch & bound
algorithm which is established on the IUR designed. Further increase the query process, they proposed
enhanced alternative of the IUR-tree known as the cluster IUR-tree and both analogous optimization algorithm
[4].
Enhanced Text clustering IUR-tree
In the reprocessing stage, aggregate each database objects are grouped as C1… Cn as indicated by
their content similitudes. A Cluster IUR-tree (CIUR-tree) expand each IUR-tree hub by the group data to create
a cross breed tree. CIUR-tree is manufactured in light of the spatial vicinity as done in the IUR-tree.
Nonetheless, every hub of a CIUR-tree incorporates another passage as (ID: N) ClusterList, where ID is the
group id and N is the quantity of objects of bunch ID in the hub of a subtree. The layer of ClusterList above C
Parent overlaps the lower layer C Child i.e, C Parent.
N = CChildj. N𝑀
𝑗=1 --------- (4)
Here, the no of children in a tnode is M [4].
Another system for productive nearest neighbor search in an arrangement of high-dimensional focuses.
The system depends on the pre-calculation of the arrangement space of any subjective closest neighbor. Relates
the calculation of the information focuses of the Voronoi cells. As Voronoi cells may turn out to be fairly
perplexing while going to higher measurements, another calculation as the guess of Voronoi cells high-
dimensional utilizing an arrangement is least bouncing rectangles. In spite of the fact that strategy depends on a
pre-calculation arrangement area, it is alterable, i.e., it underpins and inserts new information focuses. At last
appeared in an exploratory assessment that method is effective for different sorts of information and
unmistakably outflanks the best in class nearest neighbor algorithms [1].
Def 1. (Voronoi Cell, Voronoi Diagram)
Voronoi Cell(A) = {x𝜖𝑅𝑑|∀ 𝑝𝑖 ∈ 𝐴 ∀ 𝑝𝑗 ∈ 𝐷𝐵𝐴 : 𝑑(𝑥, 𝑝𝑖)} ---------(5)
The order m Voronoi figure of the DB is described as
Voronoi Diagramm (DB) = (VoronoiCell (A)|A⊂DB∧|A|=m} ----------(6)
Order 1 of a Voronoi Cells, also known as NN-cells (Fig.1b)
(a) (b)
Fig.1 Voronoi and Nearest Neighbor
Def 2. Nearest Neighbor cell, Nearest Neighbor figure
For each point with a distance operation, the Nearest Neighbor cell of P is described as
Nearest Neighbor cell (P) = d:Rd
× Rd
R+
0 ----------- (7)
{x∈DS| ∀ (p∈DB{P}):d(x,P) ≤ d(x,P)} ---------------------- (8)

NN figure, db of the points DB is given as
(DB) = {NN Cell(P) | P∈DB} ----------------(9)
As mentioned in Def 2, the addition of amount of each Nearest Neighbor Cells is the figure of the data space :
𝑎𝑚𝑡(𝑁𝑁 −𝑁
𝑖=1 Celli ) = amt(DS) -------------(10)
Def 3. (MBR approximates of Nearest Neighbor cells)
The APPRMBR of a NNC is the minimum bounding rectangle
MBR=(l1,h1,......,ld,hd) of NNC, i.e. for I = 1,......,d:
Li = min{Pi|P∈ NNC} and hi = max{pi|p∈ NNC}-------(11)
Fig. 2: NN-cells and their MBR-approximations
Outlined an algorithm to process the briefest ways between all the vertices in the system and utilizing a
most limited way quad tree to catch spatial rationality. With the calculation, the briefest ways between all
conceivable vertices can be figured to answer different kNN questions on a given spatial system. In any case, all
the aforementioned procedures, essentially centered on the separation metric. Content depiction (keyword) of
spatial items in their question assessment procedures are not considered [6].
The aggregate spatial keyword query, introduced the new issue of recovering a gathering of spatial
items, and each connected with an arrangement of catchphrases. Estimation calculations with provable guess
limits and correct calculations to tackle the two issues are produced [3]. Consider S, an arrangement of
keywords. Catchphrases might catch client inclinations or utilized venture accomplice abilities, contingent upon
the application. Give D a chance may include in a database comprising of m spatial objects. Every article D in o
is connected among an area o.λ & an arrangement of catchphrases o.ψ, o.ψ ⊂ S, that portray the item. Assume
spatial gathering catchphrase question q = ⟨q.λ, q.ψ⟩, where q.λ is an area and q.ψ speaks to an arrangement of
watchwords. The spatial gathering watchword inquiry finds a gathering of items χ, χ ⊆ D, such that the cost
Cost(χ) and such that ∪r∈_r.ψ ⊇ q.ψ is minimized. To present cost capacities give an arrangement of items χ,
the cost capacity has both weighted segments:
Cost(q, χ) = αC1(q, χ) + (1 − α)C2(χ)---------- (12)
where C1(・) is subject to the separation of the items in χ to the inquiry article and C2 (・) describes
the between item removes among the articles in χ. This sort of cost capacity is fit for communicating that
outcome item ought to be close to the inquiry area (C1(・)), that the outcome articles ought to be close to each
other (C2 (・), it has two perspectives which gives distinctive weights (α). Assume instantiations of the two
cost capacity Cost(q, χ) which trust coordinates that are planned applications well.
TYPE1 cost capacity:
Cost(q, χ) =Σr∈_(Dist(r, q)) ------------ (13)

The cost capacity is the addition of the distance among every object in χ with the query address. Which
can fit for the system of an object is used to fit the query address, therefore approaching or the searching of a
system.
TYPE2 cost capacity:
Cost(q, χ) = α maxr∈_(Dist(r, q)) + (2)(1 − α) max r1;r2∈_(Dist(r1, r2))-------- (14)
An efficient method for top-K spatial query is proposing an indexed IR2-tree which combines
signature data and R-tree to confess keyword research for spatial points that have finite no of keywords. Using
IR2-tree an effective increment algorithm is implemented to state the spatial quering keyword [10].
SI-Indexing is produced that is an index method that may solute the keyword with a query
point. SI-index method in minute seconds that decreases the computation cost and produce the solution also
with this, the secure area position algorithm by the answer rotating area query with key is produced. To answer
queries like unified spatial key queries, query with distinctive extent of continuous or spatial keyword with
query or many exact locations may be found in seconds with association of distinct methods [19].
Exact algorithm and Approximate algorithm, IR-Tree: This method is used to extract a set of keywords
and objects which are closest to the keyword location by a set of spatial web points where the minimum similar
object distances have the keywords concealed. Method addresses instantiation that group keyword query. First
is to search the set of objects that enclose the words with the addition of their distances to the keyword is lower.
Second is to search a set of objects that covers the query which adds the maximal distance among an object in a
set of objects and distance among two objects in a set of objects is lowest. Both of these sub problems are NP-
complete. Greedy algorithm is used to provide an approximation answer to the issue that uses the spatial query
indexing Information Retrival-tree to reduce the search area. But in some application query does not contain a
large number of keywords, for this exact algorithm is used that uses the dynamic programming [12].
A methodology that figures the importance between the article and a question of the reports. This
pertinence is then consolidated with the Euclidean separation in the middle of article and inquiry to ascertain a
general closeness of the item to question [11]. A area mindful top-k text retrieval (LkT) question recovers
database D in k objects for a given inquiry Q where as their areas are nearest to the area determined in Q and
their printed depictions are the most pertinent to the keyword in Q. Here inquiry Q = (loc, keywords) where
Q.loc is an area descriptor and Q.keywords is an arrangement of watchwords, the articles returned are
positioned by positioning capacity f(Dε, P(Q.keywords|O.doc)), where Dε is the Euclidian separation in the
middle of Q and O and P(Q.keywords|O.doc) is the likelihood of producing question Q.keywords from the
dialect models of the reports, which will be utilized to rank the items. In particular, given a question Q and a
record O.doc, the positioning capacity for the inquiry probability dialect model is as per the following:
P(Q.keywords|O.doc) = t∈Q.keywords ∏ pˆ(t|θO.doc)----------(15)
Geometric properties in meeting a query for searching the nearest neighbor along with text play are a
key role. The special features with documents or signature files of the latest SI-Index approach. It is believed
that an exhaustive list of Nearest Neighbor Search as reported is provided [13].
Approximation Algorithm
The primary NP-finished is a small problem by lessening from the Weighted Set Cover (WSC). The
lessening in the confirmation is guess safeguarding. Along these lines, the guess properties of the WSC issue
persist the issue. For the issue of WSC, Hk-estimate calculation for the weighted k-set spread it is realized that a
voracious calculation, where Hk = ∑k i=1 1 i is the k is the quantity of inquiry keywords, k-th consonant
number issue. In this way, we can adjust the eager calculation to prepare the spatial gathering catchphrase
question. The key of the line is the expense of every component and as the order and levels are shown in the
below Table No: 4.
Figure out the cost of an object o is by
Dist (q,o)
|0.ψ∩qs.ψ|
; ----------- (16)
Figure out the cost of a node entry e is by
minDist (q,e)
|e.ψ∩qs.ψ|
; ---- (17)
here minDist(q, e) views the minimum distance among q & e.

Table No: 4 Approximation Levels
An effective index(distance signature) for separation calculation and question handling over long
separations are proposed. A strategy discretizes the separations in the middle of items and system hubs into
classes and afterward encodes these classifications to execute the kNN look process, with a specific end goal to
accelerate kNN search [5].
Fig.3 (a) MBRs overlaying R-tree (b) Signatures of entries
The introduction of hybrid indexing structure bR-tree, that associates the bitmap ratio and R-tree to
generate m-nearest keyword inquiry arrival the spatial nearest objects identical keywords m. They utilized a
priori based search strategy that successfully reduces the search space and also proposed two monotone
constraints, distance mutex and keyword mutex to help effective pruning [9].
A framework for spatial keyword query that can method GIR system and concentrate on categorization
methods. It offers a framework for Geo- graphic data Retrieval (GIR) Systems in query generation. Develop a
unique categorization structure referred to as KR*-tree that taking the collective delivery of keywords in area
and far increase act over prasent index structures. Practically to the present solutions on actual GIS datasets
display the efficient methods are correlated [21].
III. Conclusion
Contextual search that is integrated with efficiency supports novel varieties of abstracting queries are
seen for calling a search engine in many applications. The current answers to the queries may acquire
precaution area uses or are inadequate to produce actual solutions. Now days it is important to know the
hospital’s specialties and the services that are offered from them. The work uses contexts as diseases a query

and searches for the hospitals that offers the given disease and displays the specialized hospital with its distance.
It uses the following approaches remedied the situations by, semantic based search along with ontology and a
spatial inverted index. Semantic approach assumes several points, counting context search of location. SI-index,
still has the capacity to achieve NN search in few seconds of time but also fairly space economical.
References
[1] Rosslin, John Robles, Fast Nearest-Neighbor Search Algorithms Based on High Multidimensional Data, Asia-Pacific
Journal of Multimedia Services Convergence with Art, Humanities and Sociology, Vol.3, No.1, 2013, pp. 17-24.
[2] K.Shiva Krishna, Fast Nearest Neighbor Search With Keywords, International journal of Computer Science and Mobile
computing. Vol.3 Issue.9, 2015, pg.340-350.
[3] Cao, G. Cong, C.S. Jensen, and B.C. Ooi, Collective Spatial Keyword Querying, Proc. ACM SIGMOD Int’l Conf. Management of
Data, 2011, pp. 373-384.
[4] J. Lu, Y. Lu, and G. Cong, Reverse Spatial and Textual k Nearest Neighbor Search, Proc. ACM SIGMOD Int’l Conf. Management
of Data, 2011, pp. 349-360.
[5] Hu et al, An Efficient index, International journal of Computer Science and Mobile computing, Vol.3 Issue.9, 2014 pg. 340-
350.
[6] Sam et al, An algorithm to compute shortest path, International journal of Computer Science and Mobile computing, Vol.3
Issue.9, 2014, pg. 340-350.
[7] Yufei Tao and Cheng Sheng, Fast Neighbor Search with Keywords, IEEE Transactions on Knowledge and Data
Engineering, Vol.26, No.4, April 2014.
[8] Pawar Anitha R, Designing of Semantic Nearest Neighbor Search, An international journal of advanced computer technology,
Volume-IV, Issue- II, 2015.
[9] D. Zhang, Y.M. Chee, A. Mondal, A.K.H. Tung, and M. Kitsuregawa, Keyword Search in Spatial Databases: Towards Searching by
Document, Proc. Int’l Conf. Data Eng. (ICDE), 2009, pp. 688-699.
[10] I.D. Felipe, V. Hristidis, and N. Rishe, Keyword Search on Spatial Databases, Proc. Int’l Conf. Data Eng. (ICDE), 2008.65. IEEE,
2008, pp. 656-665.
[11] G. Cong, C.S. Jensen, and D. Wu, Efficient Retrieval of the Top-k Most Relevant Spatial Web Objects, PVLDB, vol. 2, no. 1, 2009,
pp. 337- 348.
[12] Anjum Zareen and Saktel Priti, Survey on Nearest Neighbor Search for Spatial Database, International Journal of
Computer Science and Information Technologies, Vol. 5, No. 6, 2014, pp. 7101-7103.
[13] C. Usha Rani and N. Munisankar, Spatial Index Keyword Search in Multidimensional Database, International Journal of
Computer Science and Information Technologies, Vol. 5, No. 5, 6468-6471, ISSN: 0975-9646, 2014.
[14] Cristiano Rocha, Daniel Schwabe and Marcus Poggi de Aragao, A Hybrid Approach for Searching in the Semantic Web,
Proceedings of the 13th
international conference on World Wide Web, 2004, pages 374-383.
[15] Darshan Bhansali and Harsh Desai, A Study of Different Ranking Approaches for Semantic Search, International Journal of
Computer Applications (0975 – 8887), Volume 129 – No.5, November 2015.
[16] Delroy Cameron, Amit P. Sheth, Nishita Jaykumar, Krishnaprasad Thirunarayan, Gaurish Anand, and Gary A. Smith, A Hybrid
Approach to Finding Relevant Social Media Content for Complex Domain Specific Information Needs, PMC, Dec 1-2015.
[17] K. R. Uthayan and G. S. Anandha Mala, Hybrid Ontology for Semantic information Retrieval Model using keyword Matching
indexing System, The Scientific World Journal, Article ID 414910, 2015, 9 pages.
[18] Fausto Giunchiglia, Mikalai Yatskevich, Pavel Shvaiko, Semantic Matching: Algorithms and Implementation, Journal on Data
Semantics IX, Volume 4601 of the series Lecture Notes in Computer Science, 2007, pp 1-38.
[19] Komal K. Chhajed, Single and Multiple point Spatial Queries Supporting Keywords for Searching Nearest Neighbors, International
Journal of Computer Applications (0975 – 8887) Volume 110 – No. 7, January 2015.
[20] SayaliBorse, Nearest Neighbour Search With Keywords In Spatial Databases, International Journal of Innovative Research in
Science, Engineering and Technology, Vol. 4, Issue 11, November 2015.
[21] Minojini N, Gayathri R Krishna, Rekha A, Sowmiya A P, Dynamic Nearest Neighbor Search With Keywords, International
Journal of Advanced Research in Computer and Communication Engineering Vol. 4, Issue 3, March 2015.

G1803054653

Related slideshows

More Related Content

G1803054653