SEMANTIC INFORMATION EXTRACTION IN UNIVERSITY DOMAIN

Natarajan Meghanathan, et al. (Eds): SIPM, FCST, ITCA, WSE, ACSIT, CS & IT 06, pp. 513–527, 2012.
© CS & IT-CSCP 2012 DOI : 10.5121/csit.2012.2350
SEMANTIC INFORMATION EXTRACTION IN
UNIVERSITY DOMAIN
Swathi , Tamizhamudhu , Sandhiya Devi
Department of Information and Technology,College of Engineering,Guindy,
Anna University,Chennai-25
rsswathi12@gmail.com,t.amudhu@gmail.com,d27sandhya@yahoo.com
ABSTRACT:
Today’s conventional search engines hardly do provide the essential content relevant to the
user’s search query. This is because the context and semantics of the request made by the user
is not analyzed to the full extent. So here the need for a semantic web search arises. SWS is
upcoming in the area of web search which combines Natural Language Processing and
Artificial Intelligence.
The objective of the work done here is to design, develop and implement a semantic search
engine- SIEU(Semantic Information Extraction in University Domain) confined to the
university domain. SIEU uses ontology as a knowledge base for the information retrieval
process. It is not just a mere keyword search. It is one layer above what Google or any other
search engines retrieve by analyzing just the keywords. Here the query is analyzed both
syntactically and semantically.
The developed system retrieves the web results more relevant to the user query through keyword
expansion. The results obtained here will be accurate enough to satisfy the request made by the
user. The level of accuracy will be enhanced since the query is analyzed semantically. The
system will be of great use to the developers and researchers who work on web. The Google
results are re-ranked and optimized for providing the relevant links. For ranking an algorithm
has been applied which fetches more apt results for the user query.
KEYWORDS:
Natural Language Processing, Semantic Analysis, Ontology, Web Filtering
1.INTRODUCTION:
Conventional web search engines are the most widely used system nowadays for searching and
retrieving the results. But the problem is that, the documents and contents are retrieved only
based on keywords. This may not provide the most relevant and useful content related to the user
query. Here semantics of the query is not considered. It is a mere keyword based search.
A user may also require the web services associated with the retrieved content. But the generic
search engines do not provide the web services associated with the request automatically. If the

514 Computer Science & Information Technology ( CS & IT )
query is a location dependent query the apt results relevant to the location may not be retrieved
appropriately.
1.1 SEMANTIC WEB
The Semantic Web is a Web with a meaning. It describes things in a way that computers can
understand. It is an extension to the normal Web and is not about links -relationships between
things and its properties. Conventional Web consists of human operator and uses computer
systems for tasks like finding, searching and aggregating whereas Semantic Web is the one
understood by computers, does the searching, aggregating and combining information without a
human operator. It is easily processable by machines, on a global scale. It is the efficient way of
representing data on the World Wide Web.
1.1.1 LIMITATIONS IN CONVENTIONAL WEB SEARCH
Conventional web search engines are the most widely used system nowadays for searching and
retrieving the results. But the problem is that, the documents and contents are retrieved only
based on keywords. This may not provide the most relevant and useful content related to the user
query. Here semantics of the query is not considered. It is a mere keyword based search.
A user may also require the web services associated with the retrieved content. But the generic
search engines do not provide the web services associated with the request automatically. If the
query is a location dependent query the apt results relevant to the location may not be retrieved
appropriately.
1.1.2 NEED FOR SEMANTIC WEB:
The above limitations present in the conventional web search is overcome by building a semantic
search engine, thereby analyzing the meaning of the query and providing more appropriate results
to the users through keyword expansion.
1.2 OBJECTIVE
The aim of the project is to design and implement a semantic search that retrieves the search
results analyzing the context and semantics of the query. The semantic search retrieves the most
relevant results for the queries under university domain. This search is made possible by
construction of a strong ontology which forms the knowledge base. The system eliminates the
irrelevant results by forming refined queries and ranking the retrieved links.
1.3 PROBLEM DEFINITION
In this information age, it is a deplorable state that despite the overload of information, we
regularly fail to locate relevant information. Particularly, in the field of education, several
terabytes of content related to various educational institutions such as universities, colleges are
uploaded on the internet every week, and the demand for such resources is always on the rise. But
access to this information using a generic search engine is not satisfactory in terms of the
relevance of links and the overtime on bad links. This can be attributed to several factors, the

Computer Science & Information Technology ( CS & IT ) 515
most important being the absence of identification of context and semantics of the user query in
fetching the required results.
In order to overcome these critical issues the proposed system Semantic Information Extraction
in University Domain(SIEU) is designed. SIEU retrieves the semantically relevant results for the
user query by considering the semantics and context of the query. The Semantics of the query is
analyzed by means of the following procedures:
• The user query is initially analyzed grammatically and syntactically by parsing.
• The related synsets for the keywords in the query are retrieved.
• The domain related keywords in the ontology are retrieved to form the refined
query.
The results obtained in SIEU are more relevant by adopting the following procedure
• The refined queries that serve as the input for the search engine are formed based
on the semantic analysis of the user query.
• The web links retrieved for all the newly formed refined queries are re-ranked
based on the domain specific information.
In this way SIEU provides a semantic search that retrieves the appropriate results for the user
query.
1.4 SCOPE
• Provides an exclusive search service for university related information on the web
• People belonging to different domain can retrieve university related information in
an easier way.
• The web results are ranked and more appropriate to the user query.
• The users of this system are provided with the satisfactory results.
This system can also be implemented in a mobile device
2. LITERATURE SURVEY
Semantic Web Searches are an upcoming trend in WWW search. They let knowledge workers
concert their efforts and provide a high degree of relevancy and accuracy. Semantic information
extraction can be achieved through a multitude of approaches. In this section, we present a survey
of some of the existing systems and highlight their unique features.
2.1 EXISTING SYSTEMS
Semantic Information Retrieval has become the core part of any search engine. Many papers deal
with SWS that uses the OWL language for constructing ontology. DySE System (Dynamic
Semantic Engine) [1] implements a context-driven approach in which the keywords are processed
in the context of the information in which they are retrieved, in order to solve semantic ambiguity
and to give a more accurate retrieval based on user interests. DySE splits the user query into

subject keywords and the domain specific keywords. It uses a dynamic system that constructs
ontology dynamically and uses that as a knowledge base. The DSN (Dynamic Semantic Network)
is created by the DSN Builder, which generates it from Word Net by means of the domain
keyword submitted by the user during query submission. In this way a relevance assessment is
made in order to compare the results. Ontology Construction in Education Domain [2] deals with
the construction of Ontology for specific University constructing instances specifically. Here the
usage of Protégé tool for constructing the ontology is illustrated. It states the various issues which
play a key role in realizing the vision of semantic web such as XML (Extensible Markup
Language) and XML Schema, RDF(Resource Description Framework and RDF Schema,
URI(Uniform Resource Identifier), Unicode and SPARQL(Standard Protocol for RDF Query
language), Search Engines and Agents, and Ontology etc. Ontology development is the objective
of the above system and it has provided the guidelines to work in it with education domain as
example. Query sentences as semantic networks [3] paper describes procedure for representing
the queries in natural language as semantic networks. Here a syntactic analysis of the query is
done by parsing the query using Stanford parser to tag each and every word with their
corresponding parts of speech. Candidate set generation is an important method used here. A
plain text based and word net based comparisons are done to match the related concepts in the
ontology. There are several ways to information from ontology. Semantic Information Retrieval
System [4] is mainly concerned with retrieving information from a sports ontology using the
SPARQL query language. Here specific information is retrieved from the ontology. The sports
related information is queried from the ontology and it is done using SPARQL language. It
provides a basement for any further research to achieve intelligent fuzzy retrieval of sport
information through fuzzy ontology. The pages retrieved from web search needs to be ranked for
getting more relevant links. A Relation-Based Page Rank Algorithm for Semantic Web Search
Engines [5] proves that relations among concepts embedded into semantic annotations can be
effectively exploited to define a ranking strategy for Semantic Web search engines. This sort of
ranking behaves at an inner level (that is, it exploits more precise information that can be made
available within a Web page) and can be used in conjunction with other established ranking
strategies to further improve the accuracy of query results. With respect to other ranking
strategies for the Semantic Web, this approach only relies on the knowledge of the user query, the
web pages to be ranked, and the underlying ontology. Thus, it allows one to effectively manage
the search space and to reduce the complexity associated with the ranking task.
The overview of the existing systems gives multitude approaches for semantic information
extraction. Though these above systems perform a semantic analysis, it has been implemented in
a more generic way. Hence in order to further enrich this process to retrieve more promising
results a system has been proposed for queries relating to university domain (SIEU).In this
proposed system, in combination with some of the above said methodologies, some more
procedures have also been added to perform semantic information extraction in a better way.
2.2 CONTRIBUTIONS
SIEU has been designed for retrieving promising results for the queries under university domain.
Here after performing a syntactic and semantic analysis of the user query, with the keywords
extracted from ontology the refined queries are formed and sent to the search engine. Once the
web links are obtained after sending it to a conventional search engine, a re-ranking algorithm has
been proposed which justifies that the most relevant web links are filtered and ranked with higher
importance and then the less relevant links are produced. Our system also classifies that if it is a

location based user query, an analysis is done based on certain keywords in the input query and is
separately processed for fetching the most apt results for what the user has asked.
3. INITIAL PROCEEDINGS
Ontology is “a formal explicit specification of a shared conceptualization”. Ontology provides a
common understanding of a term and also its relationship with other terms. Thus a hierarchy can
be formed with the related terms. Thus considering our domain, each University will express their
purposes and functionalities in different terms. The user query should be parsed so that the stop
words that are not needed can be removed from the query. By parsing the query the nouns, verbs
and other parts of speech can be used separately if needed. Also Word Net can be used to get
synsets of the verbs so that the query can be refined more. The nouns can be used to get their
related terms and properties from the constructed ontology.
4. FLOW DIAGRAM

5. DETAILED DESCRIPTION
5.1 ONTOLOGY CONSTRUCTION: Knowledge Base
ONTOLOGY, a formal representation of knowledge as a set of concepts within a domain forms
the knowledge base for our project that is constructed based on the concepts related to the
university domain. By referring through various university websites a handful of information is
gathered and based on that a strong ontology is constructed taking into consideration, the various
important areas under university domain.
TOOL USED: Protégé 4.1
5.2 USER INPUT:
The user of the system enters a query related to university domain in natural language. The
expected output of this query is the semantically relevant web links. The irrelevant links are
filtered out.
5.3 PARSING OF INPUT QUERY:
The input query given by the user is initially parsed by means of the parser. The parsing is done
to analyze the query syntactically which determines the part of speech of each and every word in
the query. In this way the given query is analyzed grammatically.
TOOL USED: Stanford Parser
5.4 WORDNET:
The output obtained from the parser is sent to the wordnet to get the related synsets of various
words contained in the query. So here semantically related words are obtained from the output of
the wordnet.
TOOL USED: Wordnet API
5.5 EXTRACTION FROM ONTOLOGY:
This process involves of more importance where the information related to the given user query is
extracted from the built ontology. The initially given query after passing through the Stanford
parser and wordnet a set of classified and semantically analyzed words are obtained. These words
are matched with the concepts contained in the ontology to get a set of more related key words.
At the end of this process we get a collection of words which are semantically related and domain
specific key words.
TOOL USED: Jena API
5.6 FORMATION OF REFINED QUERY:
Next process involves of query formation with these collection of words. Permutations and
combinations are needed to form various refined queries from the words obtained. The queries
formed will be more refined and will fetch more semantically related web links on passing these

queries as input to the search engines. The refined queries are sent to search API which fetches
the web links related to the user query.
TOOL USED: Google Search API
5.7 RANKING OF WEB PAGES:
The web links obtained after passing the refined queries to the search API are now filtered and
ranked to make it more refined. If any of the web links are not relevant to the given query they
are filtered out. Ranking is applied to all web links obtained from all possible queries formed
under permutation. On applying ranking the web links are re-ranked in the appropriate order of
semantic relatedness.
TOOL USED: Ranking Algorithm
6. RESULTS OBTAINED IN EVERY STAGE:
6.1 QUERY ENTRY BY THE USER:
The user enters the query in the interface provided by our system.

6.2 PARSING OF THE INPUT QUERY:
The query given by the user is parsed by means of Stanford parser and the output is:
6.3 RETRIEVAL OF SYNSETS FROM WORDNET:
Now the related synsets for the words present in the query are retrieved from the wordnet.
6.4 EXTRACTION OF DOMAIN KEYWORDS FROM ONTOLOGY:
The domain keywords that are semantically related to the words in the query are extracted from
ontology.

6.5 WEB LINKS RETRIEVED:
6.5.1 User Query:
6.5.2 With Refined Query:

7. PERFORMANCE STUDIES
7.1 MEASURES USED
Recall:
Measure of how much relevant information the system has extracted (coverage of
system).
The Recall calculated here is the relative recall in which performance is compared in
relative to the google search engine.
Precision:
Measure of how much of the information the system returns is correct
(accuracy).
7.2 SIEU VS GOOGLE
When our proposed system SIEU was tested, a marked improvement in performance was
observed for most of the queries as a result of semantic analysis, although a small fraction of
them had negative and similar performance with a generic search engine.
The table 5.11 shows the samples of the performance levels of our system comparing it with
Google.
Table 1. Precision and Recall values of sample queries
SAMPLE QUERIES GOOGLE SIEU
PRECISION RECALL PRECISION RECALL
colleges for doing M.B.A 0.68 0.44 0.87 0.5
teaching staff in computer
science department in Anna
university
0.62 0.41 0.86 0.6
RECALL = # of relevant links given by the system
Total # of relevant links in GOOGLE and SIEU
PRECISION = # of relevant links given by the system
Total # of links retrieved

professors with more
number of publications in
IIT in department IT
0.68 0.5 0.78 0.54
last date to apply for M.S in
Stanford university
0.56 0.43 0.77 0.57
financial aid offered for
summer internships in UK
0.75 0.46 0.87 0.56
Deadline for payment of fees
for M.B.A course in sastra
university
0.53 0.31 0.77 0.56
Associations formed for
students in California
university
0.7 0.45 0.88 0.55
Provide me the details of the
chairman of board of
committee members
0.66 0.52 0.73 0.6
Research areas in IIT where
foreign collaborations exists
0.56 0.5 0.68 0.54
Details about the facilities
available in research
institutions of delhi
university
0.6 0.55 0.57 0.55
Provide me the information
about the correspondence
students of MIT
0.7 0.45 0.7 0.56
Road maps to visit the
campus of Stanford
0.65 0.55 0.78 0.61

university
Information regarding the
universities in abroad which
provides internship in
accounting
0.68 0.46 0.60 0.45
Procedure to apply online
for M.S in U.S university
0.74 0.45 0.79 0.65
How far is tagore university
located from anna nagar
0.76 0.58 0.81 0.59
What are colleges located
near by tambaram for doing
regular M.E course
0.67 0.45 0.83 0.55
The above table depicts that the precision value of our system SIEU is higher than the values
obtained in google search engines. The relative recall values estimates the retrieval effectiveness
between Google and our system. The more relative recall values of out system shows that SIEU is
more effective in retrieval than Google search engine.
7.3 Precision Recall curve
Figure 5.1 shows the precision Vs recall graph for SIEU and GOOGLE system .The graph is
drawn taking the first 5 queries into consideration, their corresponding precision and recall values
are plotted for both Google and SIEU systems. The precision recall curve in the graph clearly
depicts that SIEU system retrieves the accurate links for the user query based on semantic
relatedness. The values of precision and recall depicts the performance of SIEU system.

Figure 1. Precision Vs Recall graph for SIEU Vs GOOGLE
Table 2. Average precision and recall of SIEU
SIEU GOOGLE
AVERAGE
PRECISION
0.79 0.64
AVERAGE RECALL 0.55 0.48
Table 5.12 shows the average precision and recall of SIEU and Google. From this we can infer
that the higher value of average precision and recall for our proposed system SIEU when
compared to Google depicts that our system has a better performance and accuracy in retrieving
the results than the generic search engines. The recall values depict the coverage of the system.
The average relative recall value of SIEU system is also higher compared to that of the Google.
The higher value denotes the best coverage of our system compared to the generic search engines.
8. CONCLUSION:
Semantic relevant information has been retrieved as a result of this system. To add on to it many
more services are to be added. Location based information retrieval is an additional feature
where we have planned to use google maps for this purpose by means of which our system gets
enhanced with this location independent feature. And invocation of web services may be provided

to the users if incase there exists a web service related to their query which could be made
possible with the help of RSS feeds. Thus we believe an information system with these enhanced
features will be developed under university domain.
9. REFERENCES:
[1] Antonio M. Rinaldi, “An Ontology-Driven Approach for Semantic Information Retrieval on the Web”
,2009, ACM Transactions on Internet Technologies, Vol .9, no.3, pp:10:1-10:24.
[2] Sanjay Kumar Malik ET. al. ,“Developing a University Ontology in Education Domain using Protégé
for Semantic Web”, 2010, International Journal of Engineering Science and Technology, Vol.2, no.9,
pp:4673-4681.
[3] Joel Booth, Barbara Di Eugenio, Isabel F. Cruz, Ouri Wolfson,” Query Sentences as Semantic (Sub)
Networks”,2009, IEEE International Conference on Semantic Computing,Chicago,USA,pp:89-92
[4] Jun Zhai, Kaitao Zhou,” Semantic Retrieval for Sports Information Based on Ontology and SPARQL”,
2010 , International Conference of Information Science and Management Engineering (ICME),Xian,Vol.19
no.2 pp: 315-323.
[5] Fabrizio Lamberti,Andrea Sanna, and Claudio Demartini, ” A Relation-Based Page Rank Algorithm for
Semantic Web Search Engines”,2009 ,IEEE Transactions on Knowledge and Data Engineering, Vol.21,
no.1,pp:123-136.
[6] Stijn Vandamme, Johannes Deleu,” CROEQS: Contemporaneous Role Ontology-based Expanded
Query Search --Implementation and Evaluation”,2009, International Conference on Communication
Software and Networks, Ghent, Belgium,Vol.22,pp:448-452.
[7] Chen-Yu Lee, Von-Wun Soo, ”Ontology based Information Retrieval and Extraction”, 2005, IEEE
Transactions on Knowledge and Data Engineering,pp:265-269.
[8] Haibo Yu, Tsunenori Mine and Makoto Amamiya,” An Architecture for Personal Semantic Web
Information Retrieval System– Integrating Web services and Web contents”, 2005 ,Proceedings of the
IEEE International Conference on Web Services, Orlando, Florida,vol.5 ,No.3,pp.329-336
[9] Regina M.M.Braga, Claudia M. L. Werner ,Marta Mattoso,” Using Ontologies for Domain Information
Retrieval”,2000, IEEE International Conference on Database and Expert Systems
Applications,Brazil,pp:836-840.

APPENDIX A
• Stanford parser
Stanford parser is a natural language parser that works out the grammatical structure of
sentences, for instance, which groups of words go together (as "phrases") and which words
are the subject or object of a verb. Tagging process is done by the parser where the words are
tagged by their parts of speech.
• Word net
Word Net is a large lexical database of English. Nouns, verbs, adjectives and adverbs are
grouped into sets of cognitive synonyms called as synsets, each expressing a distinct concept.
Java API for Word Net Searching (JAWS) is an API that provides Java applications with the
ability to retrieve data from the Word Net database
• Jena Framework Api
Jena is a Java framework for building Semantic Web applications. It provides a programmatic
environment for RDF, RDFS and OWL and includes a rule-based inference engine. Jena is
open source.
• Google search Api
Google Search Api coordinates a search across a collection of search services. It provides all
kinds of search such as local search, web search, News search, Video search etc. Google Api
loader loads this search Api which provides the search results from the generic search
engines.
• HTML Parser
HTML Parser is a Java library used to parse HTML. Primarily it is used for transformation or
extraction. It features filters, visitors, custom tags. It is a fast, robust.It is used to extract the
meta tags from the web pages.

SEMANTIC INFORMATION EXTRACTION IN UNIVERSITY DOMAIN

Recommended

More Related Content

What's hot (19)

Similar to SEMANTIC INFORMATION EXTRACTION IN UNIVERSITY DOMAIN (20)

More from cscpconf (20)

Recently uploaded (20)

SEMANTIC INFORMATION EXTRACTION IN UNIVERSITY DOMAIN