Word Sense Disambiguation

Word sense disambiguation (WSD) is the problem of determining the sense of a multi-sense word in a given context. It is one of the important processes needed for natural language processing applications. Inductive Logic Programming (ILP) is the area of artificial intelligence which contains machine learning and logic programming. It aims to build first- order theories from examples and background knowledge, which are also represented by first- order clauses. The richness of first order logic employed in ILP can hopefully provide advantages for NLP applications such as WSD. In this study, a corpus-based WSD application is improved by using ILP.

Many times people use a single word with multiple senses, which provides a different meaning based on the sentence in which it has been used. So, the main goal of the work is to disambiguate an ambiguous word that has been located in certain sentences. Thus, it helps in exploring the actual sense of the word that it means. This process is itself named as Word Sense Disambiguation (WSD) which is an essential and ongoing subject in NLP. However, the role of domain is very helpful in exploring the actual sense of an ambiguous word. There are a number of approaches to WSD which takes a wider semantic space of ambiguous words into account [10]. This semantic space can be represented as a specific domain, task or application [10]. The domain information which is one among them is more advantageous in the process of disambiguation, thus, the work explores the role of the domain in the disambiguation process. The work also includes a score allotment to all the senses of an ambiguous word based on the semantic relation of it with the ambiguous word. Later, this helps in obtaining the actual sense of the word. Keywords: Disambiguation, natural language processing, word sense I. INTRODUCTION In NLP, ambiguity has become a barrier to human language understanding. The best solution to overcome this barrier is the Word Sense Disambiguation process. For instance, consider an example word 'bank' which has two senses. One sense of the word is "financial reservoir" and another sense of it is "a river edge". Now consider the following two statements. "Willows lined the bank of the stream" and "I went to the bank to get a home loan" To a human, the difference between these two senses of the word 'bank' will be clearly understandable as it means 'river bank' in the first sentence and a 'financial reservoir' in the second sentence. But this is not the same in case of a machine. Thus, it needs a different solution. There are different solutions of WSD that have been proposed. This can be generally divided into supervised and knowledge-based approaches. In comparison of these two approaches, the knowledge-based approach has gained a rapid development while compared to others in recent years [12]. Also, the availability of abundant information from a different knowledge resources has narrowed the gap between these two approaches [12]. Hence, the use of a knowledge-based approach has been considered as a useful one in our work. The main idea of this approach is to make use of WordNet and a semantic space such as specific domains to a greater extent to obtain the actual sense of the ambiguous word. Thus, the main purpose of our work is to make use of the domain information as a best possible way to explore the actual meaning of an ambiguous word that has been used in a sentence. There is a hypothesis that the domain information provides a powerful way to establish a semantic relation among the word senses [11]. Thus, it can be used in a profitable way in the process of disambiguation. We can refer to the domain, as a set of words that has a strong semantic relation between them. Thus, an approach of using domain information in obtaining the actual sense of an ambiguous word makes sense. Here, the basic prediction that we assumed to achieve the goal was to make use of a distinct score allotment for each of the senses that comes under the semantic space of domain of ambiguous word. This score allotment is not a random one, rather the sense that is closer to the word in a particular sentence will be allotted with a highest score while the remaining senses will be allotted in a decreasing manner accordingly. This prediction gave a way to attain the result later. Also, this approach helps in less time consumption than the normal disambiguation process. Therefore, the efficiency of the project will also be increased. II. LITERATURE SURVEY S.G Kolte and S.G Bhirud approaches is to Word sense disambiguation using wordnet domains, here they used wordnet as database to define domains and the words in the given sentence is taken as parameter which helps to detect domain by domain-oriented text analysis. For this they go through unsupervised method, and they trying to disambiguate nouns first using pos tag and getting results but here the drawback is that this approach is failed for a word having more than one sense in a same domain [1]. A Fully Unsupervised Word sense disambiguation system using dependency knowledge on specific domain is proposed in the year 2010, here they developed a fully unsupervised system using domain specific knowledge and this system performs above the first sense baseline. they showed that wsd can be achieve in unsupervised method without get compromised with supervised approach by using easily available unannotated text from internet and other sources and get a good result [2].

Compositional-distributional models of meaning aim to unify the two prominent semantic paradigms in natural language: the compositional perspective of formal semantics and the distributional models based on vector spaces. In this way, one can start from word vectors based on co-occurrence statistics collected by a corpus, and produce a vector that represents the meaning of a sentence. However, most of the current methods aiming to providing compositionality are based on ambiguous vector representations, where all the meanings of a polysemous word, such as "bank", are fused into the same vector or tensor. For many of the compositional models, this practice produces results that are hard to interpret. In this talk we investigate the effect of Word Sense Disambiguation in the context of compositionality in vector spaces, and we provide experimental evidence that explicitly separating the two tasks in distinct steps can provide better composite representations for a number of models.

Traditional approaches to word sense disambiguation (WSD) rest on the assumption that there exists a single, unambiguous communicative intention underlying every word in a document.  However, writers sometimes intend for a word to be interpreted as simultaneously carrying multiple distinct meanings.  This deliberate use of lexical ambiguity—i.e., punning—is a particularly common source of humour.  In this paper we describe how traditional, language-agnostic WSD approaches can be adapted to "disambiguate" puns, or rather to identify their double meanings.  We evaluate several such approaches on a manually sense-annotated corpus of English puns and obse

This paper presents a new model of WordNet that is used to disambiguate the correct sense of polysemy word based on the clue words. The related words for each sense of a polysemy word as well as single sense word are referred to as the clue words. The conventional WordNet organizes nouns, verbs, adjectives and adverbs together into sets of synonyms called synsets each expressing a different concept. In contrast to the structure of WordNet, we developed a new model of WordNet that organizes the different senses of polysemy words as well as the single sense words based on the clue words. These clue words for each sense of a polysemy word as well as for single sense word are used to disambiguate the correct meaning of the polysemy word in the given context using knowledge based Word Sense Disambiguation (WSD) algorithms. The clue word can be a noun, verb, adjective or adverb.

Student obligations imply writing a large number of homework assignments and term papers. Usually, they are submitted in electronic form. Checking papers for plagiarism isn’t an easy task. Quantity prevents teachers and professors to check all of them by hand. Therefore, there is a need for a system that will perform this task automatically. This paper describes principles behind one such system. Text contained in papers written by students, but also ones that can be found on internet, is converted in n-gram models, which are kept, and used later for comparison with newly generated ones. Potential application of this system is at the Faculty of Electronic Engineering in Niš, Department of Computer Science, where it can be used to check student papers, written in Serbian language.

Ambiguity is an age old problem in translation. The homonymy, homography, and polysemy causes problem while resorting to translate the source language into target language. It is context which gives information regarding the correct sense of a word. A word is mostly polysemous denoting different meanings depending upon the context. So we are left with context for the correct interpretation of an ambiguous utterance. If we don't have the needed contextual information, the translation fails. There are many linguistic methods to resolve ambiguity for the correct selection of the sense. Nearby words, collocation, co-occurrence and the selectional restrictions by subject, object and indirect object are the linguistic methods of resolving ambiguity. But in the case of machine translation, the job is not as easy as in the case of translation by a human translator, because, the linguistic rules are difficult to inculcate to the computer. Somehow we have to artificially make the computer intelligent enough to do the translation. It is a challenging task. Here in this paper, we plan to make an outline of the problem at hand.

There has been a great impact of technology on Society. Ever since the civilisation, there has been a visible impact of technology on every society. Each new invention affects how people relate to one another and how cultures have expanded or ended. Technologies have impacts on people, culture, cities and countries. Especially with the emergence of web technologies, people have become more aware than ever and their social status has changed. It is the web and internet technology that has made a new class of rich people. It has totally changed the environment of teaching and learning. Almost everything is now on line and it has revolutionised the way people interact with each other. In this article we review the overall impact of technology specially the web and communication technology on the Indian society and its future.

This thesis aims to augment the Geographic Information Retrieval process with information extracted from world knowledge. This aim is approached from three directions: classifying world knowledge, disambiguating placenames and modelling users. Geographic information is becoming ubiquitous across the Internet, with a significant proportion of web documents and web searches containing geographic entities, and the proliferation of Internet enabled mobile devices. Traditional information retrieval treats these geographic ...

Natural Language Processing (NLP) is a very popular and research area of computer science. NLP is a part of Artificial Intelligent but NLP has combination of many fields such as Hindi, English, and Computer Science etc. This paper contains how verb work in Hindi and English languages and morphology of both languages. Morphological Analyzer and generator is a tool for analyzing the given word and generator for generating word given the stem and its features. There are many Indian languages and many European languages but generally Hindi language consider as an Indian language and English as a European language. Both languages have grammatical rules. In English language, we do not use verbs as gender identification but in Hindi we use verbs for gender identification.

My thesis aims to augment the Geographic Information Retrieval process with information extracted from world knowledge. This aim is approached from three directions: classifying world knowledge, disambiguating placenames and modelling users. Geographic information is becoming ubiquitous across the Internet, with a significant proportion of web documents and web searches containing geographic entities, and the proliferation of Internet enabled mobile devices. Traditional information retrieval treats these geographic entities in the same way as any other textual data. In this thesis I augment the retrieval process with geographic information, and show how methods built upon world knowledge outperform methods based on heuristic rules. The source of world knowledge used is Wikipedia. Wikipedia has become a phenomenon of the Internet age and needs little introduction. As a linked corpus of semi-structured data, it is unsurpassed. Two approaches to mining information from Wikipedia are rigorously explored: initially I classify Wikipedia articles into broad categories; this is followed by much finer classification where Wikipedia articles are disambiguated as specific locations. The thesis concludes with the proposal of the Steinberg hypothesis: By analysing a range of wikipedias in different languages I demonstrate that a localised view of the world is ubiquitous and inherently part of human nature. All people perceive closer places as larger and more important than distant ones. The core contributions of mythesis are in the areas of extracting information from Wikipedia, supervised placename disambiguation, and providing a quantitative model for how people view the world. The findings clearly have a direct impact for applications such as geographically aware search engines, but in a broader context documents can be automatically annotated with machine readable meta-data and dialogue enhanced with a model of how people view the world. This will reduce ambiguity and confusion in dialogue between people or computers.

Word Sense Disambiguation (WSD) is one of the most challenging problems in the research area of natural language processing. To find the correct sense of the word in a particular context is called Word Sense Disambiguation. As a human, we can get a correct sense of the word given in the sentence because of word knowledge of that particular natural language, but it is not an easy task for the machine to disambiguate the word. Developing any WSD system, it required sense repository and sense dictionary. It is very costly and time-consuming to build these resources. Many foreign languages have available these resources, that is why most of the foreign languages like English, German, Spanish etc lot of work is done in these Natural languages. When we look for Indian languages like Hindi, Marathi, Bengali etc. very less work is done. The reason behind this is resource-scarcity. In this paper, we majorly focus on Marathi Language Word Sense Disambiguation because of very less work is done...

By combining an Ancient Greek semantic domains database with a corpus of annotated Ancient Greek treebanks, one may observe semantic preferences of individual words or word combinations. This information may then be applied to phrases which contain a word of ambiguous sense with the possibility of disambiguating that word. The method was applied to John 1:5, 1 Corinthians 11:15, and the phrase εν τοις επουρανιοις which occurs throughout Ephesians. Due to the small size of the annotated Ancient Greek corpus and the limited capability of the existing Ancient Greek semantic domains database, disambiguation was inconclusive for the first two passages. In Ephesians, however, this method combined with some manual inspection demonstrated that επουρανιοις may very well be referring to heavenly beings rather than a heavenly place. This is a valuable insight which had previously been unanimously rejected by published Ephesians works on the basis of flawed logic and with no consideration of linguistic evidence.

Word Sense Disambiguation (WSD) is defined as the task of finding the correct sense of a word in a specific context. This is crucial for applications like Machine Translation and Information Extraction. While the work on automatic WSD for English is voluminous, to our knowledge, this is the first attempt for an Indian language at automatic WSD. We make use

SENSEVAL was the first open, community-based evaluation exercise for Word Sense Disambiguation programs. It took place in the summer of 1998, with tasks for English, French and Italian. There were participating systems from 23 research groups. This special issue is an account of the exercise. In addition to describing the contents of the volume, this introduction considers how the exercise has shed light on some general questions about word senses and evaluation.

Most compositional-distributional models of meaning are based on ambiguous vector representations, where all the senses of a word are fused into the same vector. This paper provides evidence that the addition of a vector disambiguation step prior to the actual composition would be beneficial to the whole process, producing better composite representations. Furthermore, we relate this issue with the current evaluation practice, showing that  disambiguation-based tasks cannot reliably assess the quality of composition. Using a word sense disambiguation scheme based on the generic procedure of Schutze (1998), we first provide a proof of concept for the necessity of separating disambiguation from composition. Then we demonstrate the benefits of an "unambiguous" system on a composition-only task.

This paper presents an adaptation of Lesk’s dictionarybased word sense disambiguation algorithm. Rather than using a standard dictionary as the source of glosses for our approach, the lexical database WordNet is employed. This provides a rich hierarchy of semantic relations that our algorithm can exploit. This method is evaluated using the English lexical sample data from the Senseval-2 word sense disambiguation exercise, and attains an overall accuracy of 32%. This represents a significant improvement over the 16% and 23% accuracy attained by variations of the Lesk algorithm used as benchmarks during the SENSEVAL-2 comparative exercise among word sense disambiguation systems.

Sentiment classification is an ongoing field and interesting area of research because of its application in various fields collecting review from people about products and social and political events through the web. Currently, Sentiment Analysis concentrates for subjective statements or on subjectivity and overlook objective statements which carry sentiment(s). During the sentiment classification more challenging problem are faced due to the ambiguous sense of words, negation words and intensifier. Due to its importance the correct sense of target word is extracted and determined for which the similarity arise in WordNet Glosses. This paper presents a survey covering the techniques and methods in sentiment analysis and challenges appear in the field.

Word sense disambiguation (WSD) is a process of identifying proper meaning of words that may have multiple meanings. It is regarded as one of the most challenging problems in the field of natural language processing (NLP). Nepali Language also has words that have multiple meanings, thus giving rise to the problem of WSD in it. In this paper, we investigate the impact of NLP resources like morphology analyzer (MA) and machine readable dictionary (MRD) in ambiguity resolution. Our results show that the accuracy in WSD is better with the availability of NLP resources like morph analyzer, MRD etc. Lesk algorithm has been used to solve WSD problem using a sample Nepali WordNet containing few sets of Nepali nouns and the system is able to disambiguate these nouns only. The system was tested on a small set of data with limited number of nouns. The accuracy reading was between 50% - 70% depending on the sample data provided. When the same data was tested through manual morph analysis, the accuracy was seen to be considerably high (80%).

This thesis contributes to ongoing research related to the categorical compositional model for natural language of Coecke, Sadrzadeh and Clark in three ways: Firstly, I propose a concrete instantiation of the abstract framework based on Frobenius algebras (joint work with Sadrzadeh). The theory improves shortcomings of previous proposals, extends the coverage of the language, and is supported by experimental work that improves existing results. The proposed framework describes a new class of compositional models that find intuitive interpretations for a number of linguistic phenomena. Secondly, I propose and evaluate in practice a new compositional methodology which explicitly deals with the different levels of lexical ambiguity (joint work with Pulman). A concrete algorithm is presented, based on the separation of vector disambiguation from composition in an explicit prior step. Extensive experimental work shows that the proposed methodology indeed results in more accurate composite representations for the framework of Coecke et al. in particular and every other class of compositional models in general. As a last contribution, I formalize the explicit treatment of lexical ambiguity in the context of the categorical framework by resorting to categorical quantum mechanics (joint work with Coecke). In the proposed extension, the concept of a distributional vector is replaced with that of a density matrix, which compactly represents a probability distribution over the potential different meanings of the specific word. Composition takes the form of quantum measurements, leading to interesting analogies between quantum physics and linguistics.

To test general and specific processes of symbol learning, 4- and 5- year-old children learned three kinds of abstract associates for novel objects: words, facts, and pictograms. To test fast mapping (i.e., one-trial learning) and subsequent learning, comprehension was tested after each of four exposures. Production was also tested, as was children’s tendency to generalize learned items to new objects in the same taxon. To test for a bias toward mutually exclu- sive associations, children learned either one-to-one or many-to- many mappings. In Experiment 1, children learned words, facts (with or without incidental novel words), or pictograms. In Exper- iment 2, children learned words or pictograms. In both of these experiments, children learned words slower than facts and picto- grams. Pictograms and facts were generalized more systematically than words, but only in Experiment 1. Children learned one-to-one mappings faster only in Experiment 2, when cognitive load was increased. In Experiment 3, 3- and 4-year-olds were taught facts (with novel words), words, and pictograms. Children learned facts faster than words; however, they remembered all items equally well a week later. The results suggest that word learning follows non-specialized memory and associative learning processes.

This paper presents a design for rule-based machine translation system for English to Marathi language pair. The machine translation system will take input script as English sentence and parse with the help of Stanford parser. The Stanford parser will be used for main purposes on the source side processing, in the machine translation system. English to Marathi Bilingual dictionary is going to be created. The system will
take the parsed output and separate the source text word by word and searches for their corresponding target words in the bilingual dictionary. The hand coded rules are written for Marathi inflections and also reordering rules are there. After applying the reordering rules, English sentence will be syntactically reordered to suit Marathi language

Many researchers believe that certain aspects of natural language processing, such as word sense disambiguation and plan recognition in stories, constitute abductive inferences. We have been working with a specific model of abduction, called parsimonious covering, ap­ plied in diagnostic problem solving, word sense disambiguation and logical form generation in some restricted settings. Diagnostic parsimo­ nious covering has been extended into a dual- route model to account for syntactic and se­ mantic aspects of natural language. The two routes of covering are integrated by defining "open class" linguistic concepts, aid­ ing each other. The diagnostic model has dealt with sets, while the extended version, where syntactic considerations dictate word or­ der, deals with sequences of linguistic concepts. Here we briefly describe the original model and the extended version, and briefly characterize the notions of covering and different criteria of parsimony. Finally we examine...