Design and Implementation of HINSPELL -Hindi Spell

International Journal of scientific research and management (IJSRM)
||Volume||3||Issue||2||Pages|| 2058-2061||2015||
Website: ISSN (e): 2321-3418
Design and Implementation of HINSPELL -Hindi Spell Checker
using Hybrid approach
Baljeet kaur1, Harsharndeep Singh2
Department of Computer Science & Engineering,
Baba Farid College of Engineering and Technology, Bathinda, India
[email protected]
Department of Information Technology
Baba Farid College of Engineering and Technology, Bathinda, India
[email protected]
Abstract: A spell checker is an application program that flags words in a document that may not be spelled correctly. A
spell checker is a basic need of a word processor of any language. Spell checker analyzes the written text in order to
identify any misspellings and gives best correct suggestions for those misspellings. Most of work has been done in English
and Punjabi language. Hindi is the third most spoken language in the world. In This paper the design, techniques and
implementation of the Hindi spell checker is proposed. Error detection, Error correction by generating suggestions and
replacement are the main features of this system. The system detects approximately 83.2% of the errors and provides
77.9% of the correct suggestions for the misspelled words.
Keywords: Error detection, Error correction, HINSPELL, dictionary lookup, weight age algorithm, M.E.D, SMT.
1. Introduction
The ways in which the words can be meaningfully combined
is defined by the language's syntax and grammar. The actual
meaning of words and combinations of words is defined by
the language's semantics. Hindi is the official language of
India which consist 11 vowels and 33 consonants. Hindi is
also the third most spoken language in the world .Spell
checking is the process of detecting and providing correct
suggestions for misspelled words in a written text. Spell
correction is a one of the main functions of word processors,
search engines, text editors, and optical character recognition
(OCR). Error detection, suggestion generator, error correction
are three main steps in a spell checker. Error Correction is a
major issue in the language processing field. Much research
has been done in this area over the years. Before studying
about error detection and correction, it’s very important to
know how spelling errors occurs.
1.1 Types of Errors:
Techniques of error detection and correction were designed on
the basis of type of spelling errors. According to various
studies, spelling error can belong to two distinct categories:
Non-word error and Real-word error [3].
Non-word errors are those error words that cannot be found in
the dictionary. E.g. ग्यान for ज्ञान.
Typographic errors [14] categorized under non-word errors
which occur when the correct spelling of the word is known
but the word is mistyped by mistake. These errors are mostly
related to the wrong key press. For example, typing आपमान
for अपमान. Real-word errors are those error words that are
acceptable words in the dictionary but not correct according to
sentence. For example, मेरा घर उस और है (incorrect) for मेरा
घर उस ओर है (correct) और is an acceptable word in the Hindi
dictionary but it occurs as an error for ओर word. Possibility of
spelling mistakes in Hindi language increases because Hindi is
a highly confusing language. Hence Hindi spell checker is the
solution for making input text correct.
2. Proposed Work
A few work is done in Hindi spell detection and correction
field and it is not an easy task to identify errors in Hindi text.
The spell checker systems are online available but as not
standalone applications. Some paid Hindi spell checker
software’s are also online available.
HINSPELL is a web based spell checking and correcting
application for Hindi language. HINSPELL only deals with
non-word errors. The main features of HINSPELL are large
correct database and user interactive.
2.1 implementation of HINSPELL
Two different applications are designed in HINSPELL. One is
dictionary creation tool, executed once to create the own
dictionary and second is a spell checker for Hindi language
and it is implemented in c# language. At start, user gives the
input Hindi text and the system detect the errors by looking up
for that particular word into the created Hindi dictionary and
provides the correct suggestions for that misspelled word in
the suggestion list. After that user can select the suggestion
Baljeet kaur1 IJSRM volume 3 issue 2 February, 2015 []
Page 2058
from the suggestion list and replace errors accordingly. The
final output is a corrected text without any spelling mistakes.
Figure 2.1: Architecture of HINSPELL[4]
Input Hindi text
Creation of
Error selection
Through weightage algorithm, weights are allocated to
generated suggestions. Statistical machine translation (SMT)
technique applied to give priority to suggestions with same
minimum edit distance. SMT is applied on suggestions to find
a most intended word from the list of suggestions. Minimum 3
words as an input are required for proper working of this
technique. In HINSPELL, SMT compare the input text with
paragraphs maintained into database for choosing the most
intended suggestion. Priority is assigned by replacing
suggestions with error word according to its previous and next
word. If their exact combination is found in the database
paragraphs then that suggestion is suggested as most suitable
word. For example, महाराजा रणजीत पंजाव के राजा थे. In
this sentence पंजाव word is an error word. According to user
Error Detection
The error detection process consists of detecting any spelling
errors in the input text. In HINSPELL, dictionary lookup
technique is applied for detecting errors in input text by
checking each word of input text for its presence in to the
created Hindi dictionary. If the word is found then it is a
correct word otherwise it considers as an error word and that
word will be added into Error word list.
Error correction
Error correction consists of two steps: the generation of
correct possible suggestions for the error word and the ranking
of suggestions [3]. Weightage algorithm, minimum edit
distance and statistical machine translation techniques are used
for error correction in HINSPELL. Minimum edit distance
(M.E.D) applied on error word to generate possible
suggestions for that word. In the process of basic editing
operations i.e. Insertion, deletion and Substitution, M.E.D
changes an error word into the possible correct word. Distance
between error word and dictionary words are measured. The
dictionary word having minimum distance with error word is
ranked higher in suggestion list. Table 2.1 shows the possible
suggestions and minimum edit distance (M.E.D) of some error
it may be possible that correct word will पंजाब or पंजा. SMT
will give priority to suggestion by making word combinations
like [रणजीत पंजाब के] and [रणजीत पंजा के]. The word
combination which will be found into database that suggestion
will be most intended suggestion. Most intended suggestion
will be arranged on the top of the suggestion list by applying
Bubble sort algorithm.
Figure 2.2: Flow chart of hybrid approach used in
Enter Hindi Text
Dictionary lookup
Found word in
Table 2.1: Minimum Edit Distance (M.E.D)
Error word
Operation & performed
Deletion (ाा) (M.E.D=1)
substitution स (M.E.D=1)
Insertion (ाा) (M.E.D=1)
Substitution ल (M.E.D=1)
Substitute (M.E.D=1)
Deletion ाु , insert ाी
Baljeet kaur IJSRM volume 3 issue 2 February, 2015 []
Add word to Error word list
Apply Edit Distance on Error word to
& generate suggestions
Allot weightage algorithm to the
Apply SMT on suggested words to
increase weight age of most intended
Apply Bubble sort algorithm
Replacement input with most
intended word
Page 2059
2.2 Outlook of HINSPELL
Figure 2.3: shows the user interface of HINSPELL. User gives
the input and click on spell check button, error words will be
shown into error word list. When the User will select the error
word from the error word list; possible suggestions will be
shown into suggestion list.
If the most intended suggestion found in the suggestion list
then the user will replace the error word by selecting most
intended word from the suggestion list. Similarly users can
perform tasks like Reset; dictionary creation etc. virtual Hindi
keyboard can be used for typing the text. Hindi text can be
typed in any format like Devanagari, Dogri script etc.
In this research, 870 misspelled words randomly collected
from books, newspapers and peoples etc as input to test the
system. In the result analysis there are 724 words detected as
error words and system generates correct suggestions for 678
words. Hence detection rate of the system reaches 83.2%
approximately and correction rate of the system reaches 77.9%
approximately. Accuracy depends on the length of the
characters and no. of editing operations required to change an
error word into correct word.
Table 3.1. Results of HINSPELL
Detected as
an error
word in
Here D and L denote dataset and length of character
4. Conclusion and Future Scope
Figure 2.3 User Interface of HINSPELL
2.2 Dictionary creation
Dictionary creation is a tool used in spell checker application
to create the dictionary. This dictionary will be used as a
database for the spell checker. Microsoft access 2007 is used
to create a database for HINSPELL. As Shown in figure 2.4
by clicking on insert data button, words will be added into
database of the spell checker.
Figure 2.4: Dictionary creation tool
This paper presents the HINSPELL-Hindi spell checker
system which is not a part of any word processor or website.
This system only deals with non word errors. Real word errors
are subject of future research. The system gives the
approximately 83.2% detection rate and 77.9% Correction
rate. After applying SMT Technique, the accuracy of the
system increases but response time of the system also
increases so there is a scope of improvement in
implementation of SMT with less response time. HINSPELL
can also be used for other languages with modification of
dictionary and keyboard.
[1] Shikha kabra, Ritika Agarwal, February (2014) “Auto spell
suggestion for high quality speech synthesis in Hindi”,
international journal of computer applications, volume87no17.
[2] Ritika Mishra, Navjot Kaur, August (2013), “Design and
Implementation of Online Punjabi Spell Checker Based on
Dynamic Programming”, International Journal of
Advanced Research in Computer Science and Software
Engineering, Volume 3, Issue 8.
[3] Neha Gupta, Pratistha Mathur, December (2012), “Spell
Checking Techniques in NLP”: A Survey, International
Journal of Advanced Research in Computer Science and
Software Engineering Volume 2, Issue 12.
3. Results Analysis
Baljeet kaur1 IJSRM volume 3 issue 2 February, 2015 []
Page 2060
[4] Rupinderdeep Kaur, Parteek Bhatia, May (2010) “Design
and Implementation of SUDHAAR-Punjabi Spell
Checker”, International Journal of Information and
Telecommunication Technology, vol.1, Issue 15.
[5] Mrs. Namrata Tapaswi, April (2012), Dr. Suresh Jain, Mrs.
Vaishali chourey, “morphological-based spell checker for
Sanskrit sentences”, international journal of scientific &
technology research volume1, issue 3,
[6] Amit Sharma & Pulkit Jain, April (2013) “Hindi Spell
Checker”, Indian Institute of Technology Kanpur”.
[7] Li Zhao, (2009) “Based on the Phonetic Spelling
Correction System Research and Implementation”, Xi’an
Technological University, Xi’an, china, IEEE.
[8] S. Dasgupta, C.H. Papadimitriou and U.V. Vazirani,
/ vazirani/ algorithms.html
[9] Gurpreet Singh Lehal, (2007) “Design and Implementation
of Punjabi Spell Checker”, International Journal of
Systemic, Cybemetics and Informatics, pp.70-75.
[10]G S Lehal & Meenu Bhagat, “Spelling Error Pattern
Analysis of Punjabi Typed Text”, In Proceedings of the
2007 International Symposum on Machine Translation,
NLP and TSS, pp. 128-141.
[11]Francesco Bonchi, Ophir Frieder, Franco Maria Nardini,
Fabrizio Silvestri and Hossein Vahabi, (2012) “Interactive
and Context-Aware Tag Spell Check and Correction”
[12]Suzan Verberne, (2002) “Context-sensitive spell checking
based on word trigram probabilities”.
[13]Youssef Bassil & Mohammad Alwani May (2012),
“Context-sensitive Spelling Correction using Google Web
IT 5-Gram Information,” Department of Computer and
Information Science, Vol. 5, No.3.
[14]F.J. Damerau, (1964), “A Technique for Error Detection
and Correction of Spelling Errors”, Communication ACM,
pp. 171-176.
[15]Monisha Das, S. Borgohain, JuliGogoi, S. B. Nair,
“Design and Implementation of a Spell Checker for
Assamese”, in proceedings of the (2002) Language
Engineering Conference (LEC’02), pp. 156.
[16]R.E. Gorin, (1971) “SPELL: A spelling checking and
correction program”, Online documentation for the DEC10 computer.
[17]Ritu aggrawal, September (2007), “Hindi editor with spell
checker”, Vinayaka Mission University, Salem.
[18]Peterson James (1980) “Computer Programs for Detecting
and Correcting Spelling Errors”, Computing Practices
Communications of the ACM.
[19]Tanveer Siddiqui, U.S.Tiwary (2008), “Natural Language
Processing and Information Retrieval” Oxford university
[20]Prof.Puspak Bhattcharya and Prof. Rushikersh Josh,
“Design and implementation of morphology based
spellcheckers for Marathi”, TDIL Newsletter.
[21]Mukand Roy, Gaur Mohan, Karunesh K Arora,
“Comparative study of spell checker algorithm for building
a generic spell checkers For Indian language C-DAC
NODIA ,India.
[22]Veena Dixit, Satish Dethe, Rushikesh K. Joshi, “Design
and Implementation of a Morphology-based Spellchecker
for Marathi, an Indian Language”, Indian Institute of
Technology Bombay, India.
[23]K. Kukich (1992) “Techniques for automatically
correcting words in text”, ACM Computing Surveys.
24(4): 377-439.
[24]Robert & Cherry, Lorinda L, March (1975), “Computer
Detection of typographic errors”, IEEE Trans Professional
Communications, vol. PC-18, no.1 pp 54-64.
[25]Ajit Kumar, vishal Goyal, “Tdil programme: a
government initiative”, Department of Computer Science,
Punjabi University, Patiala.
[26]Deepak Seth, Mieczyslaw M. Kokar2, “SSCS: A Smart
Spell Checker System Implementation Using Adaptive
Software Architecture”, Northeastern University, Boston,
MA 02115, USA.
[29] Spell guru available at
Author Profile
Baljeet Kaur is a Student of M.Tech
(computer science Engg.) at Baba Farid
College of engineering and technology,
Bathinda. She has received her B.Tech in
Computer Sciences from Baba Farid
College of engineering and technology,
Bathinda in 2012. She is persuing her
M.Tech Thesis in the area of Natural Language Processing.
Baljeet kaur1 IJSRM volume 3 issue 2 February, 2015 []
Harsharndeep Singh received M.Tech
degrees in Information Technology from
Maharishi Markandeshwar University,
Mullana, Ambala in 2012. He is working
as Assistant Professor in Department of
Information Technology at Baba Farid
College of Engineering and Technology,
Bathinda, India.
Page 2061
Baljeet kaur1 IJSRM volume 3 issue 2 February, 2015 []
Page 2062