Word Sense Disambiguation

WSD

Word Sense Disambiguation is a crucial NLP technique for determining the meaning of a word when it is used in a specific context. Correctly detecting words in NLP systems is common, and choosing the exact usage of a word in a phrase has various applications. Word Sense Disambiguation is a technique for resolving ambiguity in determining the same word's meaning in different contexts.

In computational linguistics, word-sense disambiguation (WSD) is a problem that involves determining which sense of a word is used in a phrase. The solution to this problem has ramifications for various computer-related work, including discourse, search engine relevance, anaphora resolution, coherence, and inference. Because natural language demands a reflection of neurological reality as formed by the abilities offered by the brain's neural networks, computer science has long struggled to produce natural language processing and machine learning capabilities in computers.

Many techniques have been investigated, including dictionary-based methods that use the knowledge encoded in lexical resources, supervised machine learning methods that use a corpus of manually sense-annotated examples to train a classifier for each distinct word, and completely unsupervised methods that cluster occurrences of words to infer word senses. The most successful algorithms to date have been supervised learning techniques.

Applications

WSD offers a wide range of applications in text processing and natural language processing:

• WSD can be used in conjunction with Lexicography. A lot of modern Lexicography is dependent on corpora. WSD can provide important textual indicators when employed in Lexicography.
• Text Mining and Information Extraction jobs can also benefit from WSD. Because the primary goal of WSD is to accurately identify the meaning of a word in a specific context or sentence, it can be used to classify words correctly.
• For example, a text system should be able to distinguish between a coal "mine" and a land "mine" from a security standpoint.
• While the former is helpful for the industry, the latter creates a security risk. As a result, a text-mining tool must be able to distinguish between the two.
• In the same way, WSD can be utilized for information retrieval. Text data is the primary source of information for information retrieval systems. Knowing the importance of a word in a phrase can undoubtedly assist you.

Difficulties in WSD

Differences Between Dictionaries

The most typical issue is discrepancies between dictionaries or text corpora. Different dictionaries define words differently, causing the meaning of the words to be understood differently. There is a lot of text information, and it is frequently impossible to comprehend it all efficiently.

Sense inventory and algorithms task-dependency

A task-independent sense inventory is not a logical concept:[14] each activity necessitates its categorization of word meaning into task-relevant senses. Furthermore, various applications may require completely different algorithms. The difficulty of target word selection arises in machine translation.

Discreteness of Senses

Another issue is that words cannot be broken down into individual meanings. Words often have similar meanings, which produces a lot of confusion.

Methods Of Implementing WSD

There are four main ways to implement WSD.

• Dictionary- and knowledge-based methods:

These methods rely on text data from dictionaries, thesaurus, and other sources. It is based on the notion that definitions contain terms related to one another. The well-known Lesk technique, which we'll go over later, is a seminal dictionary-based method.

• Supervised Methods:

Machine learning models are trained using sense-annotated corpora in this category. However, one issue is that creating such corpora is difficult and time-consuming.

• Semi-Supervised Methods:

Due to the lack of such a corpus, most word sense disambiguation algorithms employ semi-supervised techniques. The procedure begins with a limited amount of data, frequently created manually. This is how an initial classifier is trained. To build a more extensive training set, this classifier is applied to an untagged section of the corpus. Essentially, this method entails bootstrapping from the seed data, which is the original data.

As a result, semi-supervised methods employ both labeled and unlabeled data.

• Unsupervised Methods:

Unsupervised Methods are the most difficult for researchers and NLP specialists to master. These models assume that similar meanings and senses occur in similar contexts. They are not reliant on human labor. Hence they can break the knowledge acquisition stalemate.

Implementation

A classic Word Sense Disambiguation algorithm is the Lesk Algorithm. The Lesk method assumes that words in a specific text region will have similar meanings. The correct meaning of each word context is obtained using the Simplified Lesk Algorithm by finding the sense that overlaps the most between the supplied context and its dictionary definition.

Importing Libraries
from nltk.wsd import lesk
from nltk.tokenize import word_tokenize
Downloading necessary libraries
import nltk
nltk.download('wordnet')
Let’s try with example one:
a1= lesk(word_tokenize('On Saturdays, the banks will not accept cash'),'bank')
print(a1,a1.definition())
a2 = lesk(word_tokenize('The river burst its banks'),'banks')
print(a2,a2.definition())
Output
Synset('savings_bank.n.02') a container (usually with a slot in the top) for keeping money at home
Synset('bank.n.10') a flight maneuver; aircraft tips laterally about its longitudinal axis (especially in turning)
Take another example:
a1= lesk(word_tokenize('This device is used to jam the signal'),'jam')
print(a1,a1.definition())
a2 = lesk(word_tokenize('I am stuck in a traffic jam'),'jam')
print(a2,a2.definition())

You can also try this code with Online Python Compiler

Run Code

Output

Synset('jamming.n.01') is deliberate radiation or reflection of electromagnetic energy to disrupt enemy use of electronic devices or systems.
Synset('jam.v.05') get stuck and immobilized

That's the basic implementation of the WSD.

FAQs

1. What isn't required for deciphering word senses?
The issue with WSD is that multiple algorithms may be required for different applications. For example, it takes the form of target word selection in machine translation, and there is no need for a sense inventory in information retrieval.

2. What is the purpose of word sense disambiguation?
Word sense disambiguation (WSD) is a challenge in natural language processing that involves detecting which its use in a particular context activates "sense" (meaning) of a word. This process appears to be primarily unconscious in individuals.

3. Where does word sense disambiguation come into play?
WSD offers a wide range of applications in text processing and natural language processing. WSD can be used in conjunction with Lexicography. A lot of modern Lexicography is dependent on corpora. WSD can provide important textual indicators when employed in Lexicography.

Key Takeaways

Let us brief out the article.

Firstly, we saw the basic introduction of WSD and the relevance of WSD. Later, we have a detailed discussion about WSD with some of its applications. Moving on, we saw the difficulties faced in WSD. Last, we saw the different basic methods to implement WSD. Lastly, we saw the basic implementation of WSD.

Also check out - :

• Data Dictionary In Software Engineering
• Coding Jam Google

I hope you all like this article. Want to learn more about Data Analysis? Here is an excellent course that can guide you in learning. Can also refer to our Machine Learning course.

Happy Learning Ninjas!