Text Mining in Digital Libraries using OKAPI BM25 Model

International Journal of Computer Applications Technology and Research
Volume 7–Issue 10, 398-406, 2018, ISSN:-2319–8656
www.ijcat.com 398
Text Mining in Digital Libraries using OKAPI BM25 Model
Gesare Asnath Tinega1
Student SCIT,
JKUAT
Nairobi, Kenya
Prof. Waweru Mwangi2
Associate Professor SCIT,
JKUAT
Nairobi, Kenya
Dr. Richard Rimiru3
,
Senior Lecturer SCIT, JKUAT
Nairobi, Kenya
Abstract: The emergence of the internet has made vast amounts of information available and easily accessible online. As a result,
most libraries have digitized their content in order to remain relevant to their users and to keep pace with the advancement of the
internet. However, these digital libraries have been criticized for using inefficient information retrieval models that do not perform
relevance ranking to the retrieved results. This paper proposed the use of OKAPI BM25 model in text mining so as means of
improving relevance ranking of digital libraries. Okapi BM25 model was selected because it is a probability-based relevance ranking
algorithm. A case study research was conducted and the model design was based on information retrieval processes. The performance
of Boolean, vector space, and Okapi BM25 models was compared for data retrieval. Relevant ranked documents were retrieved and
displayed at the OPAC framework search page. The results revealed that Okapi BM 25 outperformed Boolean model and Vector Space
model. Therefore, this paper proposes the use of Okapi BM25 model to reward terms according to their relative frequencies in a
document so as to improve the performance of text mining in digital libraries.
Keywords: Online Public Access Catalogs, Relevance Ranking, Digital Libraries, Okapi BM25 Model, Text Mining, Information
Retrieval Models
1. INTRODUCTION
The internet and information technology evolution has
drastically transformed information development and access,
especially in the library sector thus disrupting the
functionality of libraries. As a result, majority of the libraries
have digitized their content in order to remain relevant and
exist in distributed networks [11]; [7]. Users are now using
Public Access Catalogs (OPAC) to search and retrieve
information from the digital library’s database [5]. Khiste,
Deshmukh & Awate [8] defined digital libraries as huge
collection of electronic information that can be accessed by
distributed users from different locations. In their study
Dwivedi; Sharma & Patel, defined OPAC as a library catalog
that displays a large collection of materials held by a database
in which users search to access the desired documents
available at a library by using in search terms such as the
author, title, subject/keyword, or date of publications of the
material [5]; [17].
However, studies reveal that digital libraries are still losing to
other online search engines such as Amazon despite the
efforts to transform library catalogs from traditional card
cataloging to digital cataloging using Open Public Access
Catalogs (OPACs). This is so because the results retrieved at
the library's OPAC catalog does not satisfy the users need.
Kumar & Vohra [9] explains that the majority of OPACs
requires exact search terms to perform relevancy ranking
otherwise they will display the 'no output/null retrieval in the
results section. Others simply rank the results using last
in/first out. The most cataloged items will show up ending up
not meeting the expectations of the user. The digital libraries’
OPAC use the Boolean model for information retrieval which
retrieves too many or too little of the documents. These causes
havoc to users when searching relevant results. It is therefore
in the interest of the researcher, to establish how to improve
search capabilities in the digital libraries by implementing the
Okapi BM25 algorithm in order to improve relevance ranking
in the online public access catalogs (OPACs) before the
results are displayed to the user. The Okapi BM25 model is
based on the term frequency, length normalization to improve
the relevance performance of the digital libraries especially
during retrieval.
2. LITERATURE REVIEW
2.1 Digitization
Information and communication technology (ICT) in libraries
and many organizations has led to the increase of soft data
and digitization of materials [10]. Materials are digitalized to
improve their online accessibility, sorting, transmission and
retrieval. Digitization refers to the process of converting print
media to the digital content for electronic storage, access, and
distribution among users [3]. The digitization process has
facilitated storage and enhanced ease manipulation of the
traditionally digitized content by researchers [25]. The process
has further decentralized information storage therefore
making information in the digital libraries readily accessible
from anywhere anytime around the globe.
2.1.1 OPAC catalog
Online public access catalog is one of the most important
tools that contain all the bibliographic collection of documents
stored in the digital library database [19]. The frequent use of
the internet among the researchers has slowed the usage of the
library catalogs since they lack most of web 2.0 features such
as relevancy ranking [12]. The huge unstructured and
amorphous data available in the digital library databases has
on the other hand made it difficult for developers to come up
with algorithms for enhancing successful information retrieval
that matches the user queries [3]. In their study Kumar &
Vohra [9] established that 12.5 % of the library users at Guru
Nanak Dev University found the OPAC catalogue to be slow
and complicated to use thus they needed help from librarians.
Current generations of library users are not satisfied with the
results that the catalog retrieves because they display either
too many or too little documents in a given search. The recent
developments of the newer catalogs by organizations outside
of libraries have resulted in vocal criticisms about the
capability of digital libraries especially on relevance ranking
[1].

www.ijcat.com 399
2.3 Text mining
This paper adopts the definition of Talib et al [21] that defines
text mining as a type of indexing which aims at extracting
structured text data from unstructured text data. Text mining
process involves gathering, preprocessing, and text analysis of
document from various sources. These processes are carried
out to ensure user satisfaction when accessing structured data
from unstructured databases. Text mining techniques such as
information retrieval, classification, clustering and
categorization are thereafter used to ensure that data is
analyzed and generated correctly [27]. This paper will
however focus on the information retrieval (IR) approach
since it aims at retrieving relevant data to users from a large
library database.
2.4 Information Retrieval Process
The main objective of the OPAC catalog is to retrieve relevant
documents from a large library database so as to satisfy the
user information need. Information retrieval models are used
to perform the matching process between the library database
and the user query for retrieval. The three basic processes
involved in information retrieval include indexing, query
formulation and matching [13]. Indexing refers to the
document representation process. Query formulation also
known as indexing is done to by unique terms expressed by a
user while query evaluation also known as matching process
is done to estimate the level of relevance of a document to a
given query [4].
Figure 1 Information retrieval process
2.5 Information retrieval models
2.5.1 Boolean Model
It is an information retrieval model grounded on set theory to
determine the prospect of document retrieval. Boolean model
is an example of exact match model whereby the fate of the
documents retrieval is determined based on the type of
information stored in the database [14]. The model uses the
logical AND, OR, and Not operators to perform document
search in the library databases [23]. The AND operator
retrieves results that include all the keywords linked with the
operator while OR operator produces results that contain
either one or all the keywords used in the user query. The
NOT operator retrieves results that excludes the keyword
from the user query. The Boolean model is however criticized
of lack of relevance ranking when used in retrieval systems
such as the OPAC catalog. Boolean model also does not
support length normalization of the documents since it does
not use term weight such as term frequency and inverse
document frequency when retrieving documents from the
library database [2].
2.5.2 Vector Space Model (VSM)
This model was introduced to overcome the limitations of
Boolean model by assigning weights to term for better
matching. VSM presents text documents as vectors to find the
similarity between the documents stored in the database and
the user query using cosine similarity. Moreover, the model is
also used to find exact results with relevance ranking [17].
VSM obtains relevance ranking and information retrieval
using document indexing, weighting of the indexed terms
using the TF-IDF and finally ranking the documents archives
as per the query comparability value [6]. The cosine similarity
of the VSM is calculated using the equation 1 below.
Where: dj represents the
total collection of documents, q signifies the user query, Wi,j is
the ith
term of a vector for document j, Wi,q= is the ith
term of
a vector for query q, and N= is the total number of keywords
in a given data set. The model, however, faces some major
drawbacks such as poor representation of long documents
which is as a result of repetitive use of terms. Moreover, Jain,
et al [29], established that the model has low sensitivity to
semantics. For instance the word “car” and “automobile” will
not give the same match if both words are found in same
document. A study by Yulianto et al [2], also revealed that
VSM is hard to understand and takes a lot of time to search
and match documents before retrieval.
2.5.3 Okapi BM 25 model
The Okapi Best Match 25 (BM25) model is a non-binary
model that was developed as part of the Okapi Basic Search
System in the TREC Conferences. Okapi BM25 is a
probabilistic model that is based on the probabilistic theory.
The model is a well-performed term weighting scheme that
retrieves its relevant results by incorporating the use of weight
term using TF-IDF, and length normalization of a given
document [22]. BM25 is a bag-of-words retrieval function
that ranks documents according to their relevant results.
Okapi BM25 not only considers the frequency of the query
terms but also the whole the length of the document under
evaluation [26].
2.5.3.1 TF-IDF Weighting of Okapi BM25 Model
In Okapi BM25, term frequency also termed as document
frequency shows the frequency of a query term in a document
for it to be considered to be relevant. Inverse Document
Frequency (IDF), on the other hand, is used to differentiate
between common words and uncommon words within a

www.ijcat.com 400
document. The simplest score for document d can be
illustrated in the equation 2.
Where: N is the total number of documents in a given corpus;
dft is the document frequency of a term.
is an element of a query.
TF-IDF considers short documents to have more weight than
long documents therefore; Okapi BM25 model outperforms
TF-IDF and vector space model by taking the average length
of each document separately using tuning parameters. Tuning
refers to the process by which one or more parameters are
adjusted upwards or downwards to achieve an improved or
specified result. The values of the tuning parameters are
determined empirically using a test collection of documents,
queries, and relevance judgments. K1 is set to 1.2 to control
term-frequency saturation since low values result in quicker
saturation while high values results in slower saturation. The
tuning parameter b is set to 0.75 to control field-length
normalization of a document. The Okapi BM25 model
calculates the retrieval status value of a given document in
order to determine the relevance of a document as shown in
equation 3.
Where:
Retrieval Status Value: relevancy scores of a
document.
N: represents documents in a given collection.
dft-the frequency of a query term in a document.
- t is an element of query q.
t- term
q- query
tf td : signifies the frequency of a term in document d
Ld (Lave): used to calculate the average document
length in the whole collection
k1: tuning parameter set to 1.2
b: tuning parameter set to 0.75
K3 tuning parameter is set to 2 in case the retrieval involves
long documents as shown in equation 4.
2.5.3.2 Example of OKAPI BM25 Model
Example query: “president lincoln”
tfpresident,q= tflincoln,q= 1
No relevance information: R= ri= 0
“president” is in 40,000 documents in the
collection: dfpresident= 40,000
“lincoln” is in 300 documents in the collection: dflincoln=300
The document length is 90% of the average length: dl/avg(dl)
= 0.9
We pick k1= 1.2, k2= 100, b= 0.75. Hence using the Okapi
BM formula illustrated at equation 2.13 the RSV of the query
is shown in table 1 below.
Table 1 Retrieval status values of Okapi BM25
3. METHODOLOGY
3.1 Research Design
This paper used a case study research design to generate
solutions for improving information retrieval in JKUAT
library. Experimental research was also used to manipulate
variables and determine their effect on the dependent variable.
This study involves manipulation of text mining technique
such as information retrieval to improve the OPAC catalog
used in digital library.
3.2 Model Design
A prototype was used to develop this model. Prototype model
was selected because it allows development, verification in
terms of performance, and reworking on the framework until
an acceptable prototype is finally achieved. The prototype
processes help to complete a given framework in the area of
study. The figure 2 below illustrates the OPAC model design
that was used for the development of the model.

www.ijcat.com 401
Figure 2 Opac Model design
3.3 OPAC framework Requirements
The front end of the proposed OPAC catalog was
implemented using HyperText Markup Language (HTML),
Cascading Style Sheets (CSS), Bootstrap, Laravel framework,
and JavaScript language. MySQL was used to develop the
database while server side programming of the OPAC system
was done by Hypertext Preprocessor (PHP).
3.4 Document Gathering
The study utilized secondary data from the Google search
engine and other online journals. The collected document
were pre-processed using Google to remove inconsistencies
such as tokenization, stop words and stemming before the
documents were downloaded to be populated to the database.
Different search queries were used to collect all the 300
documents that were used to create the database from online
journals such as strategic journal of business and change
management, scientific research an academic publisher, and
International Journal of Computer Science and Engineering
Survey among others. For instance, the query “Text mining
and digital library” was used as a user query using the search
engine and resulted in 10 articles were displayed on the first
page of the search engine. Seven documents that were found
in Portable Document Format were collected uploaded to the
database for further analysis shown in table 2. This process
was repeated until the collection of 300 documents was
achieved.
Table 2 Document gathering
3.5 Entity Relationship Diagram.
The database_item was populated with the 300 documents
collected as shown in figure 3. An Entity Relation Diagram
(ERD) that was used to create the database.
The database is made up of two entities namely administrator
and books. The entities use one to many relationships and
therefore, one administrator or a user can add many books to
the OPAC system. The user queries the database using either
of the attributes of the book entity.

www.ijcat.com 402
Figure 3 ERD of the Opac database
3.6 Stemming Process
Stemming seeks to reduce different grammatical forms of a
word like its noun, adjective, verb, adverb among others and
remove various suffixes from a word to get its common origin
[24]. This is done to the retrieval models so as to save time
and memory space. For example, the words user, usage, using,
and usability can be rooted in the word use. The process of
stemming helps a retrieval model to have exact matching
stems and increase their performance level especially in
document retrieval. This can be illustrated in figure 4.
Figure 4 Stemming output
3.7 Routing
All the OPAC framework routes are registered within
the app/routes.php file. This file tells the php framework
(laravel) the URIs it should respond to and the associated
controller that will give it a particular call.
3.8 Search and matching process
The user uses the search box that was created at the front end
to query the database for the results to be processed by the
information retrieval models. Tokenization of the documents
is done to remove inconsistencies such as commas, full stops
among others. Matching is done before the results are
displayed to the user. It seeks to compare the user query
against the indexed documents. This result in a ranked list of
documents that will be used by the users in search of the
information they need. The Boolean Model, Vector Space
Model and the Okapi BM25 were utilized for this study.
Retrieved results were displayed on the performance basis of
each model. The fact that VSM and Okapi BM25 rank their
results qualified them to be effective ranking models as
compared to Boolean model. The following code was used for
search query and matching process
//boolean search
foreach($books as $book){
$keywords = explode(',', $book->title);
foreach($phrases as $phrase){
if(stripos(json_encode($keywords), $phrase) !== false){
if(in_array($book, $items)){
} else {
array_push($items, $book);
}
}
}
}
//vector space
$vectoritems = array();
$books = Book::all();
$keywords = explode(',', $book->abstract);
if(in_array($book, $vectoritems)){
} else {
array_push($vectoritems, $book);
} } }
//okapi
$okapiitems = array();
$books = Book::all();
$keywords = explode(',', $book->keywords);
if(in_array($book, $okapiitems)){
} else {
array_push($okapiitems, $book);
}
}
}
}
3.9 Database connection Code
The front end and the back end of the OPAC system were
connected to produce the results through the database
connection. The following PHP code was used to connect
MySQL and select item-database
<?php
$mysqli= new ,mysql (“localhost”, “username”, “ password”,
“dbname”);
?>
When the above code connects MySQL and selects item
database the user queries can now be used at the search page
to display the results.
3.10 Retrieved Relevant Documents
The improved OPAC catalog is then used to retrieve relevant
documents from the database. The retrieved relevant
documents are then displayed at the catalog for the users to
view, use and compare the performance of each information
retrieval models used.

www.ijcat.com 403
4. PERFORMANCE AND EVALUATION
OF RESULTS
4.1 OPAC Results
Once the user has installed PHP software (Xampp) in the
computer Apache and MySQL module are turned on as shown
in figure 5 below.
Figure 5 Xampp Control Panel
The user opens any browser and enters the
url:http.localhost/opac/public/users/login The following
screen will appear for the user to enter his or her email
address and a password to access the system.
Figure 6 Opac framework
The OPAC framework displays the figure below once the user
logs in the details. This can be illustrated in figure 7 below.
Figure 7 Opac framework search display
When the user hits the search button the following results are
observed as illustrated in the figure 8 below
Figure 8 Search Display
When the user searches for example the query “Information
System” the three models displayed the following results as
shown in figure 9.
Figure 9 users enters a query" information
technology
The results of the search entered by the user in figure 9 results
to the retrieval of relevant documents from each model. The
first page of the retrieved was screenshot as shown in figure
10. The Okapi BM 25 model retrieved documents by
calculating the retrieval status value of each relevant
document. Vector Space Model calculates the cosine
similarity of each document that was found to match the user
query was calculated while Boolean model retrieved just the
book title and the author name only.
Figure 10 Retrieved documents
4.2 Tests for Performance

www.ijcat.com 404
Evaluation of the OPAC information retrieval framework
performance was done and tested using precision and recall.
The Boolean model was left out because it does not retrieve
relevant results to the user. Vector space model and Okapi
BM25 were tested to proclaim the best model among the two
since their retrieval was based on relevancy. This was done
using the sample of three queries that was implied on the two
models at the same time. The improved digital library's OPAC
catalog allowed the users to search the catalog and sort the
results by relevance ranking using the three models where the
most relevant results are displayed at the top of the page.
Precision is the fraction of relevant results retrieved from the
total number of documents stored in the library database to
meet the information need of the user. Zuva & Zuka [28],
pointed out that poor performance of the models displays low
values while high performance of the models results with high
values. This can be calculated as shown in equation 5
Recall denotes the fraction of the relevant documents in the
collection returned by the system for use. This can be
calculated using the recall formula as shown in equation 6
Precision and recall calculation for query 1: Information
Systems
Table 3 Query 1- Information technology
Precision and recall for query 2: data mining in the digital
libraries today
Table 4 Query 2- data mining in the digital libraries
today
Precision and recall calculation for query 3: Challenges facing
the digital libraries especially in information retrieval
Table 5 Query 3- Challenges facing the digital
libraries especially in information retrieval
5. CONCLUSIONS
This paper’s literature review exposes a vocal dissent on the
use of OPAC in many digital libraries, especially with its
complex search mechanisms. Although recent developments
of the search capability of the OPAC have been enhanced,
still OPAC is criticized for lack of relevance ranking in its
search capability [16]. This paper concludes that Okapi BM
25 model can be used in information retrieval in the digital
library's OPAC catalogue. A term with a high relative
frequency within a document is more representative and
relevant in the document characterization and ranking. Based
on this research and analysis, the Okapi BM25 model is
proposed to reward terms according to their relative
frequencies in a document. From the results obtained, it is
clear that the Okapi BM25 model which is integrated with
relative term frequency information, document length
normalization and tuning parameters significantly
outperforms the Boolean Model and Vector Space Model on
most of the representative data collections. It is a novel
approach to combine the concept of relative term frequency
with fundamental weighting functions in probabilistic
information retrieval systems to increase performance of the
model for retrieval results in the OPAC. The OPAC
framework is accurate and applicable according to specified
requirements.
6. REFERENCES
1. Antelman, K., Lynema, E., & Pace, A. K. (2006).
Toward a Twenty-First Century Catalog.
INFORMATION TECHNOLOGY AND
LIBRARIES, 25(3), 128-139.

www.ijcat.com 405
2. Yulianto, B., Budiharto, W., & Kartowisastro, H,
I. (2017). The Performance of Boolean Retrieval
and Vector Space Model in Textual Information
Retrieval. Communication & Information
Technology, 11(1), 33-39
3. Aruleba, K. D., Akomolafe, D. T., & Afeni, B.
(2016). A Full Text Retrieval System in a Digital
Library Environment. Scientific Research
Publishing, 8(1), 1-8.
4. Boubekeur, F., & Azzoug, W. (2013).
CONCEPT-BASED INDEXING IN TEXT
INFORMATION RETRIEVAL. International
Journal of Computer Science & Information
Technology (IJCSIT), 5(1), 119-136.
5. Brahaj, A., Razum, M., & Hoxha, J. (2013).
Defining Digital Library. In T. Aalberg, C.
Papatheodorou, M. Dobreva, G. Tsakonas, & C.
J. Farrugia, Research and Advanced Technology
for Digital Libraries. (pp. 23-28). Berlin:
Springer, Berlin, Heidelberg.
6. Dwivedi, S. J. (2014). Comparative Analysis of
IDF Methods to Determine Word Relevance in
Web Document. International Journal of
Computer Science Issues, 11, 59-65.
7. Ibba, S., & Pani, F. E. (2016, May 10). Digital
Libraries: The Challenge of Integrating Instagram
with a Taxonomy for Content Management.
Future Internet, pp. 1-15.
8. Khiste, G. P., Deshmukh, R. K., & Awate, A. P.
(2018, Feb 24). LITERATURE AUDIT OF
‘DIGITAL LIBRARY’: AN OVERVIEW.
Research gate, pp. 403-411.
9. Kumar, S., & Vohra, R. (2013). "User perception
and use of OPAC: a comparison of three
universities in the Punjab region of India". The
Electronic Library, 31(1), 36-54.
10. Mishra, R. K. (2016). DIGITAL LIBRARIES:
DEFINITIONS, ISSUES, AND CHALLENGES.
Innovare Journal of Education, 4(3), 1-3.
11. O'Connell, J. (2008). Information Literacy meets
Library 2.0. In P. J, & G. P, School library 2.0:
new skills, new knowledge, new futures (pp. 51-
62). London: Facet Publishing.
12. Ogbole, J. O., & Morayo, A. (2017). Factors
Affecting Online Public Access Catalogue
Provision And Sustainable Use By
Undergraduates In Two Selected University
Libraries In Ogun And Oyo States, Nigeria.
Journal of Research & Method in Education,
7(4), 14-25.
13. Roshdi, A., & Roohparvar, A. (2015). Review:
Information Retrieval Techniques and
Applications. International Journal of Computer
Networks and Communications Security, 3(9),
373-377.
14. Ruban, s., Sam, S. B., Serrao, L. V., & Harshitha.
(2015). A Study and Analysis of Information
Retrieval Models. International Journal of
Innovative Research in Computer and
Communication Engineering, 3(7), 230-236.
15. Rybchak, Z., & Basystiuk, O. (2017). Analysis of
methods and means of text mining.
ECONTECHMOD. AN INTERNATIONAL
QUARTERLY JOURNAL, 6(2), 73-78.
16. Sankari, R. L., Chinnasamy, K.,
Balasubramanian, P., & Muthuraj, R. (2013). A
STUDY ON THE USE OF ONLINE PUBLIC
ACCESS CATALOGUE (OPAC) BY
STUDENTS AND FACULTY MEMBERS OF
UNNAMALAI INSTITUTE OF
TECHNOLOGY IN KOVILPATTI (TAMIL
NADU). International Journal of Library and
Information Studies, 3(1), 17-26.
17. Sharma, M., & Patel, R. (2013). “A survey on
information retrieval models, techniques and
applications,” International Journal of Emerging
Technology and Advanced Engineering, 3(11),
542–545.
18. Shiva Kanaujia, S., & Parveen, B. (2016).
Marketing and Building Relations in Digital
Academic Library: Overview of Central Library,
Jawaharlal Nehru University, New Delhi.
DESIDOC Journal of Library & Information
Technology, 36(3), 143-147.
19. Sundari, G. J., & Sundar, D. (2017). A Study of
Various Text Mining Techniques. International
Journal of Advanced Networking & Applications
(IJANA, 08(05), 82-85.
20. Swaminathan, K. S. (2017). Use and Awareness
of Online Public Access Catalogue (OPAC) by
Students and Faculty members of Anna
University Regional Campus, Coimbatore, Tamil
Nadu – A Case Study. International Journal of
Scientific Research and Management, 5(5), 5345-
5349.
21. Talib, R., Hanif, K. M., Ayesha, S., & Fatima, F.
(2016). Text Mining: Techniques, Applications
and Issues. International Journal of Advanced
Computer Science and Applications, 7(11), 414-
418.
22. Garcia, E. (2016, November 11). A Tutorial on
the BM25F Model - Minerazzi. Retrieved June
18, 2017, from minerazzi.com:
https://www.researchgate.net/publication/308991
534_A_Tutorial_on_the_BM25F_Model
23. Muhammad, A. B. (2017). Efficiency of Boolean
Search strings for Information Retrieval.
American Journal of Engineering Research
(AJER. 6(11), 216-222
24. Singh, V. K., & Singh, V. K. (2015). VECTOR
SPACE MODEL: AN INFORMATION

www.ijcat.com 406
RETRIEVAL. International Journal of Advanced
Engineering Research and Studies, 141-143.
25. Tuna, G., Zogo, R., & Dermirelli., B. (2013). An
Introduction to Digitization Projects Conducted
by Public Libraries: Digitization and
Optimization Techniques Journal of Balkan
Libraries Union, 1(1), 28-30.
26. Zhu, R. (2016, June 5). GRADUATE PROGRAM
IN INFORMATION SYSTEMS AND
TECHNOLOGY. IMPROVEMENT IN
PROBABILISTIC INFORMATION RETRIEVAL
MODEL REWARDING TERMS WITH HIGH
RELATIVE TERM FREQUENCY, pp. 1-95.
27. Gaikwad, S. V., Chaugule, A., & Patil, P. (2014).
Text Mining Methods and Techniques.
International Journal of Computer Applications,
85(17), 42-45.
28. Zuva, K.., & Zuva, T. (2012). Evaluation of
Information Retieval Systems. International
Journal of Computer Science & Information
Technology, 4(3), 35-43.
29. Jain, A., Jain, A., Chauhan, N., Singh, V., &
Thakur, N. (2017). Information Retrieval using
Cosine and Jaccard Similarity Measures in Vector
Space Model. International Journal of Computer
Applications , 164 (6), 28-30.

Text Mining in Digital Libraries using OKAPI BM25 Model

More Related Content

Similar to Text Mining in Digital Libraries using OKAPI BM25 Model (20)

More from Editor IJCATR (20)

Recently uploaded (20)

Text Mining in Digital Libraries using OKAPI BM25 Model