Notes

16.Differentiate between two types of representations of semantic parsing.
Ans:Types of representations of semantic parsing
-Deep Semantic Parsing-
Taking natural language input and transforming it into a meaning representation
Domain dependent
New domain – new solutions from scratch
-Shallow Semantic Parsing-
Deals with four main aspects of language:
Structural ambiguity, word sense, entity and event recognition, and predicate argument structure recognition
General purpose, low level, and intermediate representations
Extremely difficult to construct a general ontology and symbols table
Both of these approaches are fraught with issues; the first approach is so specific that porting to every new domain can require anywhere from a few modifications to almost reworking the solution from scratch. In other words, the reusability of the representation across domains is very limited.The problem with the latter approach is that it is extremely difficult to construct a general-purpose ontology and create symbols that are shallow enough to be learnable but detailed enough to be useful for all possible applications. Therefore, an application-specific translation layer between the more general representation and the more specific representation becomes necessary.

15.Explain the applications of textual entailment solutions.
ans:-Many NLP problems can be formulated in terms of recognizing textual entailment. RTE clearly has relevance to summarization, in which systems are required to generate human readable summaries of one or more documents. The subtask of identifying whether a new sentence contains information already expressed by a summary-in-progress (redundancy detection) can be thought of as an entailment pair with the present summary as the text and the new sentence as hypothesis. If T does not entail H, the sentence contains new information and should be integrated with the summary.
Information extraction comprises the task of recognizing instances of a fixed set of relations such as “works for” and “born in” in a set of natural language text documents. If we express the relations as short sentences, like A person works for an organization,” and A person was born in a location,” text spans from the source documents become the texts of entailment pairs with the reformulated relations as hypotheses, and an RTE system can be directly applied. Similarly, question answering, which requires automated systems to find candidate answers (sections of documents from a fixed document collection) to a set of questions, can be reformulated in much the same way. An RTE system can be directly applied to identify actual answers

14.Explain coverage rate and perplexity?
A:Typically, two criteria are used to define language model evaluation: coverage rate and perplexity on a held-out test set that does not form part of the training data.
The coverage rate measures the percentage of n-grams in the test set that are represented in the language model.A special case of this is the out-of-vocabulary rate (or OOV rate), which is 100 minus the unigram coverage rate, or,in other words, the percentage of unique word types not covered by the language model.
Perplexity can be thought of as the average number of equally likely successor words when transitioning from one position in the word string to the next. If the model has no predictive power at all, perplexity is equal to the vocabulary size

13.Explain the applications of sentimental analysis.
ans-Some popular sentiment analysis applications include social media monitoring, customer support management, and analyzing customer feedback. With sentiment analysis tools, however, you can automatically sort your data as and when it filters into your help desk. Let’s take a look at the most popular applications of sentiment analysis:
Social media monitoring: Social media posts often contain some of the most honest opinions about your products, services, and businesses because they’re unsolicited. With the help of sentiment analysis software, you can wade through all that data in minutes, to analyze individual emotions and overall public sentiment on every social platform.
Customer support ticket analysis: Sentiment analysis with natural language understanding (NLU) reads regular human language for meaning, emotion, tone, and more, to understand customer requests, just as a person would. You can automatically process customer support tickets, online chats, phone calls, and emails by sentiment to prioritize any urgent issues.
Brand monitoring and reputation management: Not only that, you can keep track of your brand’s image and reputation over time or at any given moment, so you can monitor your progress. Whether monitoring news stories, blogs, forums, and social media for information about your brand, you can transform this data into usable information and statistics.
Listen to voice of the customer (VoC): Combine and evaluate all of your customer feedback from the web, customer surveys, chats, call centers, and emails. Sentiment analysis allows you to categorize and structure this data to identify patterns and discover recurring topics and concerns.
Listen to voice of the employee: Use sentiment analysis to to evaluate employee surveys or analyze Glassdoor reviews, emails, Slack messages, and more..
Product analysis: You can search keywords for a particular product feature (interface, UX, functionality) and use aspect-based sentiment analysis to find only the information you need.
Market research and competitive research: Use sentiment analysis for market and competitor research. Find out who’s receiving positive mentions among your competitors, and how your marketing efforts compare

12.Write a short note on n-gram notation?
A:The probability of a word sequence W of nontrivial (!=0) length cannot be computed directly because unrestricted natural language permits an infinite number of word sequences of variable lengths.
The probability P(W) can be decomposed into a product of component probabilities according to the chain rule of probability:
Because the individual terms in this product are still too difficult to be computed directly, statistical language models make use of the n-gram approximation, which is why they are also called n-gram models.
The assumption is that all previous words except for the n − 1 words directly preceding the current word are irrelevant for predicting the current word, or, alternatively, that they are equivalent.
Depending on the length of n, we can distinguish between:
unigrams (n = 1),
bigrams (n = 2),
trigrams (n = 3), or
4-grams,
5-grams, and so on.
“John gifted a watch to his mother.”
unigrams (n = 1),
“John”, “gifted”, “a”, “watch”, “to”, “his”, “mother”
bigrams (n = 2),
“John gifted”, “gifted a”, “a watch”, “watch to”, “to his”, “his mother”
trigrams (n = 3),
“John gifted a”, “gifted a watch”, “a watch to”, “ to his mother”
So on…