<<

INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

Development of a Micro Telugu Opinion WordNet and Aligning with TELOWN Ontology for Automatic Recognition of Opinion Words from Telugu Documents

Benarji Tharini1, Dr.Vishnu Vardhan Bulusu2

1 Research Scholar Rayalaseema University, Kurnool, AP, India. [email protected] 2 Professor in Department of CSE, Manthany JNTUH, TS, India. [email protected]

Abstract: The emergencies in Indian language based documents over the web are observed in the recent past with the advent of Unicode standard. The content of Indian language and its accessibility is observed to be minimal in the linguistic process evaluation. Unicode based language tools are created in order to prepare language specific repositories and the dictionaries online. Construction of , language constructs and thinking about a semantically rich lexical synsets is useful in the linguistic processing of the Indian context. Over the two decades the research world is trending towards construction of the semantic models. It is necessary to start a beginning to create a rich knowledge base in order to attain semantically rich linguistic models. With this phenomenon as an aim a micro opinion Telugu wordnet is created in order to map with Telugu Opinion WordNet Ontology (TELOWN) which consists of semantic knowledge on positive and negative Telugu opinion words. The objective of this process is to create opinion wordnet in Telugu along with their synsets for the automatic recognition of opinion words from Telugu documents. SPARQL is used as a query language for the retrieval at the backend.

Keywords: semantic web, ontology, Telugu, WordNet, opinion words, SPARQL

I. Introduction

In a multilingual nation like India interpretation between Indian languages and also amongst English and Indian languages is a basic undertaking. Likewise basic is the errand of Cross-Lingual Search where the query is made in an Indian language and recovery of reports occurs in English or Telugu (vide Figure 1). Every one of these exercises relies upon lexical information of high caliber and scope. This lexical learning is as machine-discernable lexicons, ontologies (various leveled association of ideas) and (a huge chart like the structure of words).

Volume 7, Issue VI, JUNE/2018. Page No:197 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

Query in an Indian language IL Query in IL Query in IL

Input Input Input Processing Processing Processing

Search in Search in English Telugu

Document Document Search in Processing of Processing of IL retrieval in retrieval in English Telugu

Optional Optional

output in output in Translation Translation English Telugu E-> IL T IL

Output in IL Output in IL

Figure 1: Cross Lingual Search

The vast majority of the data on the World Wide Web is encoded as natural language content expected for people yet troublesome for machines to get it. With the Internet blast over the current years, expansive volumes of unstructured messages in different languages and structures are being included to the data stores an everyday schedule. With the approach of Unicode, this wonder is watched for writings in Indian languages like Telugu, Tamil and Bengali in the current years [1]. When all is said in done, these languages are poor as far as accessibility of entrenched corpus, natural language handling apparatuses, et cetera, and along these lines have turned into a vital zone of research in the Indian people group.

Volume 7, Issue VI, JUNE/2018. Page No:198 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

Throughout the most recent two decades, the world is seeing huge development in Web substance of Indian languages. This influenced individuals to feel good with their local language. Particularly, throughout the previous couple of years, there has been a colossal increment in the Telugu content on the web. Telugu is the fifth biggest talked language and has 250 million speakers over the world, the dominant part of who are from India [2]. So as to process the substance in local language towards important data recovery, the language- particular WordNet is required. WordNet [3] has developed as an awesome asset for the Natural Language Processing applications for English reports. Following English WordNet, WordNets are worked for some languages of the world. Indo WordNet [4] is the main WordNet worked for an Indian language.

Wordnets are lexical structures made out of synsets and semantic relations. Synsets are sets of equivalent words. They are connected by semantic relations as is hypernymy (a), meronymy (some portion of) and so on. Wordnets have developed as critical assets for Natural Language Processing (NLP). The principal word net on the planet was worked for English at Princeton University1. At that point took after word nets for European Languages: Eurowordnet2. Since 2000, wordnets for various Indian languages are getting assembled, driven by the Indo wordnet3 exertion at Indian Institute of Technology Bombay4 (IITB).

Opinionated substance in Telugu is critical to be dissected for the utilization of enterprises and government(s). Programmed thinking about such natural language records by the machine requires the help of Telugu WordNet. At the point when such a lexical asset is coordinated with the ideas of ontology, the programmed acknowledgment of opinion words from Telugu records happens effectively.

II. Related Work

A good amount of research has happened on determining orientations of the opinion words in . The development of lexical resources for both traditional information retrieval and Opinion Mining tasks is the first step in this research.

IndoWordNet is a linked lexical knowledge base of word nets of 18 scheduled languages of India, namely Assamese, Bangla, Bodo, Gujarati, TELUGU, , Kashmiri, Konkani, , Meitei (Manipuri), Marathi, Nepali, Odia, Punjabi, , Tamil, Telugu and .

Such project indeed took off in 2000 with TELUGU WordNet being created by the Natural Language Processing group at the Center for Indian Language Technology (CFILT) in the Computer Science and Engineering Department at IIT Bombay. [5] It was made publicly available in 2006 under GNU license. The TELUGU WordNet was created with support from the TDIL project of Ministry of Communication and Information Technology, India and also partially from Ministry of Human Resources Development, India.

Volume 7, Issue VI, JUNE/2018. Page No:199 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

The word nets follow the principles of minimality, coverage and replace ability for the synsets. That means, there should be at least a 'core' set of lexemes in the synsets that uniquely give the concept represented by the synsets (minimality), e.g., {house, family} standing for the concept of 'family' ("she is from a noble house"). Then the synsets should cover ALL the words representing the concept in the language (coverage), e.g., the word 'ménage' will have to appear in the 'family' synsets, albeit, towards the end of the synsets, since its usage is rare. Finally, the words towards the beginning of the synsets should be able to replace one another in reasonable amount of corpora (replace ability), e.g., 'house' and 'family' can replace each other in the sentence "she is from a noble house". IndoWordNet is highly similar to EuroWordNet. However, the pivot language is TELUGU which, of course, is linked to the English WordNet. Also typical Indian language phenomena like complex predicates and verbs are captured in IndoWordNet.

IndoWordNet is publicly brows able. The Indian language word net building efforts forming the subcomponents of IndoWordNet project are: North East WordNet project, Dravidian WordNet Project and Indradhanush project all of which are funded by the TDIL project.

Word nets of other languages of India then followed suit. The large nationwide project of building Indian language word nets was called the IndoWordNet project. IndoWordNet[1] is a linked lexical knowledge base of word nets of 18 scheduled languages of India, viz., Assamese, Bangla, Bodo, Gujarati, TELUGU, Kannada, Kashmiri, Konkani, Malayalam, Meitei, Marathi, Nepali, Oriya, Punjabi, Sanskrit, Tamil, Telugu and Urdu. The word nets are getting created by using expansion approach from the TELUGU WordNet. The TELUGU WordNet was created from first principles (mentioned below) and was the first wordnet for an Indian language. The method adopted was same as the Princeton WordNet for English.

Polish WordNet is being mapped to Princeton WordNet based on the strategy followed by IndoWordNet.[6]

For example

Fig : Telugu word response on Indo wordnet

Volume 7, Issue VI, JUNE/2018. Page No:200 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

Amitava Das and Bandopadhya created [6] SentiWordNet for the . 35,805 Bengali passages were accounted for from their trial. Joshi et al. created [7] one of Indian language Telugu SentiWordNet (T-SWN) by utilizing English SentiWordNet and English- Telugu WordNet mappings. Bakliwal et al. made [8] Telugu subjective vocabulary for Telugu content extremity grouping. They built up this dictionary with Telugu descriptive words and qualifiers and their extremity scores.

The examination chip away at connecting of WordNet with ontology is propelled by the inspiration of computerized thinking about natural language assets. The advantages of connecting WordNet with ontology are multi-overlay [10]. These are: (I) The formal details of the ontology are conceivable to be utilized with WordNet,(ii) WordNet ideas, when required, be refined and rebuilt utilizing an ontology, (iii) The formal sayings of ontology are conceivable to be connected with natural language content. This sort of connecting must prompt adroitly more thorough, subjectively straightforward and effectively exploitable WordNet in a few applications. Aldo Gangemi et al. presented [11] DOLCE upper-level ontology that situates with the semantic elucidation of WordNet scientific classification. The analysts adjusted Wordnets upper level to DOLCE ideas to make WordNet onto-coherently pleasing. Niles and Pease adjusted [12] WordNet with Suggested Upper Merged Ontology (SUMO) to answer the critical inquiry of consequently utilizing the ontology by those applications that procedure natural language content. Brijesh and Pushpak connected [10] IndoWordNet [13], a multilingual WordNet created on 17 Indian languages by the second specialist of the work with SUMO ontology.

In the above works, the local WordNets are created for all the four noteworthy word classes in particular Noun, Adjective, Verb and Adverb. To the best of creator's learning, these works never focused on building up a micro WordNet for translating solely the telugu opinion words from the electronic records as these opinion words express the data with respect to the like and abhorrence of a client on different target elements. Constant checking of different preferences data from the regularly refreshing electronic archives are exceptionally valuable to the legislatures specifically to make better administrative choices. This must happen in a programmed way as and when the application detects the approaching archives. This is conceivable to do when the WordNet is lined up with the relating ontology.

A different ontology is required to line up with the develops of micro Telugu opinion WordNet rather than standard DOLCE and SUMO ontologies. This is on the grounds that the query execution time is high [14] when these ontologies are utilized as a part of the arrangement procedure. Likewise, no complex apparatuses are accessible for separating sub- ontology [15] from upper combined ontology like SUMO which particularly focuses on natural language preparing errands. An application that backings querying the information base for exceptionally recognizing the opinion words from the Telugu reports is required.

Volume 7, Issue VI, JUNE/2018. Page No:201 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

The primary commitments of this work are: building up a Micro Telugu Opinion WordNet, building up the Telugu opinion dictionary of descriptive words utilizing Micro Telugu Opinion WordNet, creating TELOWN ontology, connecting Micro Telugu Opinion WordNet with TELOWN ontology lastly querying the TELOWN ontology to recover the opinion word related points of interest.

III. Development of Micro Telugu Opinion WordNet and Aligning with TELOWN Ontology The principal objective of linking Micro Telugu Opinion WordNet with TELOWN ontology is to make the machine automatically recognize the opinion words from the Telugu documents. In order to achieve this goal, a framework is presented in Figure 1 below.

Figure 1. Micro Telugu Opinion WordNet and TELOWN Ontology Linking and Querying

Description of Dataset The nature of the TELOWN-ed Micro Telugu Opinion WordNet is comprehended by the sort of dataset/vocabulary which thusly impacts the general execution of the Natural Language Processing (NLP) applications. Thusly, the accessibility of the datasets/dictionaries and when it isn't accessible production of reasonable datasets/vocabularies is essential for the present work. In a perfect world for content handling assignments a dataset speak to true information and contain helper information that enables mining related semantic data to make conceivable portrayals. Further, the dataset ought to be effortlessly acquirable to conduct tries different things with insignificant human endeavors.

Volume 7, Issue VI, JUNE/2018. Page No:202 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

When all is said in done, a journalistic writing is unequivocal and exact and tries to be free of language. When in doubt, writers don't use long words where short ones can mean the same. They endeavor to avoid redundancy of words for a similar section. They utilize subject-verb-protest development and dynamic composition. Words picked by the columnists are commonly in light of the subject with the purpose of featuring basic parts of the news article. News composing consolidates vocabulary and sentence structure such that the articles exhibit the data regarding relative significance of the target group. Various online news providers like AP7AM,SAKSHI,GEMINI, ETV et cetera give Telugu news articles from a wide extent of classes, for example, legislative issues, business, diversion, games, science and innovation, and so forth. The in-house dataset is gathered from the SAKSHI Telugu news site [16]. The dataset contains 1500 news articles. The insights of the SAKSHI news dataset is exhibited in Table 4 under area 4.

3.1 Pre-processing Amid pre-handling of archives in the dataset, at first, the corpus is separated into singular reports. At that point each report is separated into a rundown of words. At that point the accentuation, exceptional characters, and numbers are expelled.

In this way, the stop-words that are utilized over every one of the archives are expelled as they can't really induce any importance. This disposal depends on the stop word list gave by the University of Neuchatel [17]. At that point the copy events from the rest of the word set are evacuated leaving just extraordinary words. At long last, Part-of-Speech (PoS) labeling is done on the prevent words expelled words from the archive gathering. This PoS tagger used to label the Telugu words is Knowledge Based Computer Systems (KBCS) Telugu Postagger [18]. Just modifiers are considered for building Micro Telugu Opinion WordNet.

3.2 Creation of Micro Telugu Opinion WordNet (MTELOWN) The WordNets that are developed for native languages more or less follow the design principles of the Princeton WordNet for English, while paying particular attention to language specific phenomena whenever they arise. The work of creating Micro Telugu Opinion WordNet is easy as only Telugu adjectives for which both synonyms and antonyms are added. The principles of minimality, coverage and replace ability [13] must and should govern the creation of synsets. Synset Synset is formed by grouping the synonyms with same meaning for a word [11]. Each Synset represents one sense. Associated with each synsets of a term t, there is a definition and frequency measure that indicates the extent of the term t, is utilized in this sense.

Volume 7, Issue VI, JUNE/2018. Page No:203 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

Example 3: The frequency of “బిడ్డ [biDDa]/ శిశువు[SiSuvu]/baby” in Synsets are as follows – 1) { [biDDa] / baby పాపా [pApa]/ కూతురు [ ]/babygirl/ daughter } 2) { బిడ్డ [biDDa] / baby, బాబు [bAbu]/ కొడుకు [ ] /baby boy/son } The sense of the term “బిడ్డ [biDDa] / baby” varies based on thecontext of usage. The noun “బిడ్డ [biDDa] / baby” is more likely tobe used in the sense of పాపా “[pApa]/baby girl” or “బాబు [bAbu] /baby boy” than the sense of “కొడుకు [koDuku] /son” or “కూతురు [kUturu] / daughter”. The Synsets in WordNet are linked to each other through different relations including hyponyms, part of and member of. Whenever the sense of a given term is determined to be the Synset S, its synonyms, words or phrases from its definition, its hyponyms and compound words of the given term are considered for possible addition to the query. The similarity between words w1 and w2 can be defined as the shortest path from each sense of w1 to each sense of w2 [35]: Sim (w1, w2) = max [-log (N/2D) Where N is the no. of node in a path from w1 to w2 and D is the maximum depth of taxonomy. Sometimes the concept may belong to subsume of the concepts, in such cases the similarity can be measured as probability of the concept derived from relative frequencies of a document collection: P(c) = f(c) / N Where p(c), the probability of a concept c is defined as the ratio between frequency of a concept c and number of key terms N. The final similarity score is defined as the sum of path based similarity and content based similarity. The synsets for synonyms and antonyms are developed in this manner. Every synsets is assigned with a unique synsets ID in the alphabetical, ascending order. The purpose of this ID is to uniquely identify a common word classified under different PoS categories. English WordNet version 1.7 has 18,523 synsets [19] under adjective category. The synsets IDs of considered 100 synsets (300 positive adjective words, 400 negative adjective words) for the creation of Micro Telugu Opinion WordNet are mapped with the corresponding English WordNet synsets IDs. This mapping helps the people to understand the English equivalent word for the searched Telugu word.

Telugu Opinion Lexicon of Adjectives: Creation using Micro Telugu OpinionWordNet The considered positive and negative adjective synsets words are arranged in Telugu language dictionary order. The size of the opinion lexicon increases as and when the size of the Micro Telugu Opinion WordNet increases.

3.3 Development of Telugu Opinion WordNet (TELOWN) Ontology

Ontology characterizes an arrangement of authentic natives with which to display an area of learning or talk. The illustrative natives are ordinarily classes (or sets), qualities (or properties), and connections (or relations among class individuals).

Volume 7, Issue VI, JUNE/2018. Page No:204 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

The meanings of the authentic natives incorporate data about their significance and limitations on their coherently steady application. Ontology, together with an arrangement of individual examples of classes constitutes a learning base.

The Telugu Opinion WordNet ontology is made for Telugu opinion words, each having a positive or negative opinion. A web interface is worked to query the ontology, to restore the opinion, equivalent words, antonyms and relating English word for a given Telugu opinion word.

The TELOWN ontology is made utilizing Protégé, a mainstream open-source ontology manager. It was made in the OWL punctuation. Figure 2 portrays the structure of TELOWN ontology as a semantic system. A semantic system (likewise called idea arrange) is, where the vertices speak to ideas and where the edges speak to the relations between the ideas. This representation was made utilizing WebVOWL (Web-based Visualization of Ontologies), a web application for the intelligent perception of ontologies.

Figure 2. Semantic Network for the Telugu Opinion WordNet Ontology

Table 1 lists the classes and subclasses present in our ontology, as depicted in Figure 2. Also shown in Figure 2 are the object properties and the data properties present in our ontology. These have been listed in Table 2 and Table 3 respectively. An object property relates individuals (i.e. instances) of one class to individuals of the same class or to those of another class. A data property, on the other hand, relates individuals of a class to literals (constant values).

Volume 7, Issue VI, JUNE/2018. Page No:205 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

Table 1. Classes and Subclasses in TELOWN Ontology

S. No. Class Description

1 Word Superclass for Adjective, Adverb, Noun and 2 Adjective Verb classes 3 Adverb Subclass of Word; for opinion words that 4 Noun 5 Verb are adjectives Subclass of Word; for 6 opinion words that are adverbs Subclass of Synset Word; for opinion words that are nouns Subclass of Word; for opinion words that 7 AdjectiveSynset are verbs Super class for AdjectiveSynset, 8 AdverbSynset AdverbSynset, NounSynset and VerbSynset classes; a synsets is a collection of words 9 NounSynset having the same/similar meaning 10 VerbSynset Subclass of Synset; for adjective opinion words having the same meaning 11 Opinion Subclass of Synset; for adverb opinion words having the same meaning Subclass of Synset; for noun opinion words having the same meaning Subclass of Synset; for verb opinion words having the same meaning The opinion of a word; has 2 instances: Positive and Negative

Table 2. Object Properties in TELOWN Ontology

S. No. Object Property Domain Range 1 belongsTo Word Synset 2 hasOpinion Word Opinion 3 isSynonymOf Word Word 4 isAntonymOf Word Word

Table 3. Data Properties in TELOWN Ontology

S. No. Data Property Domain Range 1 SynsetID Synset int

3.4 `Linking Micro Telugu Opinion WordNet with TELOWN Ontology The linking of Micro Telugu Opinion WordNet with TELOWN ontology occurs by the following way:

i. getTeluguSynsetID: This process returns the synsets ID of the searched Telugu opinion word from Micro Telugu Opinion WordNet. ii. MapEnglishSynsetID: This process returns the mapping set of Telugu synsets ID and English WordNet synsets ID. This provides the English variant of the Telugu opinion word from English WordNet.

Volume 7, Issue VI, JUNE/2018. Page No:206 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

iii. alignTELOWNConcept: All the adjectives obtained after PoS tagging are annotated with the TELOWN Adjective concept as the instances. Also, the corresponding opinion orientation, synonyms, antonyms and the English Word of the searched Telugu opinion word are linked with the concept as instances. This overcomes the confusion between concept and instance.

IV. Results and Observations

The Telugu documents obtained from SAKSHI News website were used for this ex- periment. Three categories of documents were considered. Table 4 presents the details of the dataset.

Table 4. Telugu Documents Dataset Details

Document Attributes Values Number of documents in the corpus 1500 Number of categories 5 Number of documents per category 300

Once all the special characters and stop-words which do not play any role in the information processing are eliminated, the words in the document collection are PoS tagged. The sample output of PoS tagged Telugu words is displayed in Figure 6 below.

Figure 6. PoS Tagging of the Word “యుద్దము” using Knowledge Based Computer Systems Telugu PosTagger

The Telugu opinion lexicon is created with both 300 Positive and 400 Negative adjective words. The sample Telugu opinion lexicon with 16 positive adjective words and 16 negative adjective words are made available. The Micro Telugu Opinion WordNet is created with all the adjectives. Currently, 100 Telugu synsets are linked with their corresponding English synsets and TELOWN.

V. Conclusions The connecting of Micro Telugu Opinion WordNet with TELOWN makes a helpful asset for natural language preparing applications focused at the Telugu language. The formalism of TELOWN ontology can be utilized with WordNet for different content preparing applications. The framework is made accessible through TELOWN ontology querying interface that gives WordNet perusing in an all the more formally pleasant way.

Volume 7, Issue VI, JUNE/2018. Page No:207 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

This framework can be utilized for idea and connection extraction from Telugu language reports. The future point is to incorporate things, verbs, and intensifiers words and different lexical relations in and among them with descriptive words to make a Micro Telugu WordNet so the subjective words distinguished around their objectives and the objectives themselves are assessed proficiently.

References

1. Krishnamurthi, Karthik, Vijayapal Reddy Panuganti, and Vishnu Vardhan Bulusu. “Understanding Document Semantics from Summaries: A Case Study on Telugu Texts.” ACM Transactions on Asian and Low-Resource Language Information Pro- cessing (TALLIP) 16.1 (2016): 7.

2. Sharma, Richa, Shweta Nigam, and Rekha Jain. “Opinion mining in Telugu language: a survey.” arXiv preprint arXiv:1404.4935 (2014). 3. Pushpak Bhattacharyya, IndoWordNet, Lexical Resources Engineering Conference 2010 (LREC 2010), Malta, May, 2010. 4. Christiane Fellbaum (ed.), WordNet: An Electronic Lexical Database, MIT Press, 1998. 5. P. Vossen (ed.), EuroWordNet: A Multilingual Database with Lexical Semantic Networks, Kluwer Pub., 1998. 6. Joseph E. Schwartzberg,Encyclopædia Britannica, India—Linguistic Composition, 2007. 7. Dipak Narayan, Debasri Chakrabarty, Prabhakar Pande and P. Bhattacharyya An Experience in Building the Indo WordNet- a WordNet for TELUGU, International Conference on Global WordNet (GWC 02), , India, January, 2002. 8. Rudnicka, E., Maziarz, M., Piasecki, M., & Szpakowicz, S. (2012). Mapping plWordNet onto Princeton WordNet, 24th International Conference on Computational Linguistics (COLING), India, December 2012 9. Vossen, Piek. “A Multilingual Database with Lexical Semantic Networks.” Kluwer Academic Publishers, Dordrecht, 1998. 10. Das, Amitava, and Sivaji Bandyopadhyay. “SentiWordNet for Bangla.” Knowledge Sharing Event-4: Task 2 (2010). 11. http://www.wordnet.princeton.edu 12. http:// http://www.illc.uva.nl/EuroWordNet/

13. http://www.cfilt.iitb.ac.in/wordnet/webhwn

14. http://www.iitb.ac.in 15. Álvez, Javier, Paqui Lucio, and German Rigau. “Improving the Competency of First-Order Ontologies.” Proceedings of the 8th International Conference on Knowl- edge Capture. ACM, 2015. 16. Xu, Baowen, Dazhou Kang, and Jianjiang Lu. Framework of Extracting Sub- ontology.” Content Computing (2004): 493-498.

Volume 7, Issue VI, JUNE/2018. Page No:208 INTERNATIONAL JOURNAL OF RESEARCH ISSN NO : 2236-6124

17. Baek S., Cho M., Kim P. (2005) “Matching Colors with KANSEI Vocabulary Using Similarity Measure Based on WordNet.” In: Gervasi O. et al. (eds) “Computational Science and Its Applications – ICCSA 2005.” ICCSA 2005. “Lecture Notes in Com- puter Science, vol 3480.” Springer, Berlin, Heidelberg. 18. Gruber, Thomas R. “Toward Principles for the Design of Ontologies Used for Knowledge Sharing.” International Journal of Human-Computer Studies 43.5-6 (1995): 907-928. 19. Alani, Harith, et al. “Using Protégé for Automatic Ontology Instantiation.” (2004).

Volume 7, Issue VI, JUNE/2018. Page No:209