Semantic and disclosure analysis
Unit III
Sankara Narayanan S
outline
 Representing Meaning,
 Lexical Semantics,
 Word Senses,
 Relation between Senses,
 Word Sense Disambiguation,
 Word Embeddings,Word2Vec, CBOW, Skip-gram and GloVe,
 Discourse Segmentation,
 Text Coherence,
 Discourse Structure, Reference Resolution,
 Pronominal Anaphora Resolution, Coreference Resolution
Representing meaning
 Semantic analysis creates a representation of the meaning of a
sentence.
 But before getting into the concept and approaches related to
meaning representation, we need to understand the building
blocks of semantic system.
Building Blocks of Semantic System
 Entities − It represents the individual such as a particular
person, location etc. For example, Haryana. India, Ram all
are entities.
 Concepts − It represents the general category of the
individuals such as a person, city, etc.
 Relations − It represents the relationship between entities
and concept. For example, Ram is a person.
 Predicates − It represents the verb structures. For
example, semantic roles and case grammar are the examples
of predicates.
Approaches to Meaning
Representations
 First order predicate logic (FOPL)
 Semantic Nets
 Frames
 Conceptual dependency (CD)
 Rule-based architecture
 Case Grammar
 Conceptual Graphs
Lexical semantics
 The first part of semantic analysis, studying the meaning of
individual words is called lexical semantics.
 It includes words, sub-words, affixes (sub-units), compound
words and phrases also.
 All the words, sub-words, etc. are collectively called lexical
items.
 In other words, we can say that lexical semantics is the
relationship between lexical items, meaning of sentences and
syntax of sentence.
steps involved in lexical semantics
 Classification of lexical items like words, sub-words, affixes,
etc. is performed in lexical semantics.
 Decomposition of lexical items like words, sub-words,
affixes, etc. is performed in lexical semantics.
 Differences as well as similarities between various lexical
semantic structures is also analyzed.
Introduction to Semantic Analysis
 Semantic Analysis is a subfield of Natural Language
Processing (NLP) that attempts to understand the meaning of
Natural Language.
 Understanding
Natural Language might seem a
straightforward process to us as humans.
 However, due to the vast complexity and subjectivity
involved in human language, interpreting it is quite a
complicated task for machines.
 Semantic Analysis of Natural Language captures the meaning
of the given text while taking into account context, logical
structuring of sentences and grammar roles.
Parts of Semantic Analysis
Lexical Semantic Analysis:
Lexical Semantic Analysis involves understanding the
meaning of each word of the text individually.
It basically refers to fetching the dictionary meaning that a
word in the text is deputed to carry.
 Compositional Semantics Analysis: Although knowing
the meaning of each word of the text is essential, it is not
sufficient to completely understand the meaning of the text.
Semantic Analysis example
 For example, consider the following two sentences:
 Sentence 1: Students love GeeksforGeeks.
 Sentence 2: GeeksforGeeks loves Students.
 Although both these sentences 1 and 2 use the same set of
root words {student, love, geeksforgeeks}, they convey
entirely different meanings.
Tasks involved in Semantic Analysis
 Word Sense Disambiguation
 Relationship Extraction
Word Sense Disambiguation:
 In Natural Language, the meaning of a word may vary as per
its usage in sentences and the context of the text.
 Word Sense Disambiguation involves interpreting the
meaning of a word based upon the context of its occurrence
in a text.
 For example, the word ‘Bark’ may mean ‘the sound made by
a dog’ or ‘the outermost layer of a tree.’
Word Sense Disambiguation
 Likewise, the word ‘rock’ may mean ‘a stone‘ or ‘a genre of
music‘ – hence, the accurate meaning of the word is highly
dependent upon its context and usage in the text.
 Thus, the ability of a machine to overcome the ambiguity
involved in identifying the meaning of a word based on its
usage and context is called Word Sense Disambiguation.
Relationship Extraction:
 Another important task involved in Semantic Analysis is
Relationship Extracting.
 It involves firstly identifying various entities present in the
sentence and then extracting the relationships between those
entities.
Relation between senses
 Synonymy
 Antonymy
 Homonymy
 Polysemy
 Hyponymy and Hypernymy
 Meronymy and Holonymy (Part-Whole
Relationship)
Synonymy (Similar Meaning)
 Words with different forms but similar meanings.
 Example:
 Big ↔ Large
 Happy ↔ Joyful
Antonymy (Opposite Meaning)
 2. Antonymy (Opposite Meaning)
 Words with opposite meanings.
 Example:
 Hot ↔ Cold
 Fast ↔ Slow
Homonymy
 Words that have the same spelling or pronunciation but
different meanings.
 Example:
 Bank (Financial institution) vs. Bank (Riverbank)
 Bat (Animal) vs. Bat (Sports equipment)
Polysemy (Multiple Related Meanings)
 A word with multiple closely related meanings.
 Example:
 Head (Body part) vs. Head (Leader of a team)
 Run (Jogging) vs. Run (Managing a business)
Hyponymy and Hypernymy (Hierarchy)
 Hyponymy (Specific term) and Hypernymy (General
term) represent "is-a" relationships.
 Example:
 Rose → Flower (Rose is a type of Flower)
 Dog → Animal
Meronymy and Holonymy (Part-Whole
Relationship)
 Meronymy: A part of something.
 Holonymy: The whole of which something is a part.
 Example:
 Wheel → Car (Wheel is a part of a Car)
 Finger → Hand
Troponymy (Manner Relation)
 A specific manner of performing an action.
 Example:
 Whisper is a manner of Speak
 Jog is a manner of Run
Metonymy (Association-Based Meaning)
 Using a word to refer to something related to it.
 Example:
 TheWhite House (refers to the US government)
 Hollywood (refers to the film industry)
Word embedding
 Word Embedding is an approach for representing words and
documents.
 Word Embedding or Word Vector is a numeric vector input
that represents a word in a lower-dimensional space.
 It allows words with similar meanings to have a similar
representation.
Word embedding
 Word Embeddings are a method of extracting features out of text
so that we can input those features into a machine learning model
to work with text data.
 They try to preserve syntactical and semantic information. The
methods such as Bag of Words (BOW), CountVectorizer and
TFIDF rely on the word count in a sentence but do not save any
syntactical or semantic information. In these algorithms, the size
of the vector is the number of elements in the vocabulary.
 We can get a sparse matrix if most of the elements are zero. Large
input vectors will mean a huge number of weights which will
result in high computation required for training. Word
Embeddings give a solution to these problems.
Approaches for Text Representation
 Approaches for Text Representation
 The conventional method involves compiling a list of distinct
terms and giving each one a unique integer value, or id. and
after that, insert each word’s distinct id into the sentence.
 Every vocabulary word is handled as a feature in this
instance. Thus, a large vocabulary will result in an extremely
large feature size
One-Hot Encoding
 One-hot encoding is a simple method for representing words
in natural language processing (NLP).
 In this encoding scheme, each word in the vocabulary is
represented as a unique vector, where the dimensionality of
the vector is equal to the size of the vocabulary.
 The vector has all elements set to 0, except for the element
corresponding to the index of the word in the vocabulary,
which is set to 1.
Example
Disadvantages
 One-hot encoding results in high-dimensional vectors,
making it computationally expensive and memory-intensive,
especially with large vocabularies.
 It does not capture semantic relationships between words;
each word is treated as an isolated entity without considering
its meaning or context.
 It is restricted to the vocabulary seen during training, making
it unsuitable for handling out-of-vocabulary words.
Bag of Word (Bow)
 Bag-of-Words (BoW) is a text representation technique that
represents a document as an unordered set of words and
their respective frequencies.
 It discards the word order and captures the frequency of each
word in the document, creating a vector representation.
Example BOW
Example
"This is the first document.",
 "This document is the second document.",
"And this is the third one.",

"Is this the first document?“
 Vocabulary (Feature Names): ['and' 'document' 'first' 'is' 'one'
'second' 'the' 'third' 'this']
 Bag-of-Words Matrix:
[[0 1 1 1 0 0 1 0 1]
[0 2 0 1 0 1 1 0 1]
[1 0 0 1 1 0 1 1 1]
[0 1 1 1 0 0 1 0 1]]

Disadvantages
 BoW ignores the order of words in the document, leading to
a loss of sequential information and context making it less
effective for tasks where word order is crucial, such as in
natural language understanding.
 BoW representations are often sparse, with many elements
being zero resulting in increased memory requirements and
computational inefficiency, especially when dealing with
large datasets.
Term frequency-inverse document frequency (TFIDF)
 Term Frequency-Inverse Document Frequency, commonly
known as TF-IDF, is a numerical statistic that reflects the
importance of a word in a document relative to a collection
of documents (corpus).
 It is widely used in natural language processing and
information retrieval to evaluate the significance of a term
within a specific document in a larger corpus.
Term Frequency (TF)
 Term Frequency measures how often a term (word) appears
in a document. It is calculated using the formula:
 TF(t,d)=Total number of terms in document d

Total number of times term t appears in document d
Inverse Document Frequency (IDF)
 Inverse Document Frequency measures the importance of a
term across a collection of documents. It is calculated using
the formula:
 IDF(t,D)=log(Total documents /
Number of documents containing term t)
 The TF-IDF score for a term t in a document d is then given
by multiplying the TF and IDF values:
 TF−IDF(t,d,D)=TF(t,d)×IDF(t,D)
Limitation of TF-IDF
 TF-IDF treats words as independent entities and doesn’t
consider semantic relationships between them. This
limitation hinders its ability to capture contextual
information and word meanings.
 Sensitivity to Document Length: Longer documents
tend to have higher overall term frequencies, potentially
biasing TF-IDF towards longer documents.
Word2Vec
 Word2Vec is a neural approach for generating word embeddings.
 It belongs to the family of neural word embedding techniques and
specifically falls under the category of distributed representation
models.
 It is a popular technique in natural language processing (NLP) that
is used to represent words as continuous vector spaces.
 Developed by a team at Google, Word2Vec aims to capture the
semantic relationships between words by mapping them to highdimensional vectors. The underlying idea is that words with similar
meanings should have similar vector representations. In Word2Vec
every word is assigned a vector. We start with either a random
vector or one-hot vector.
Word2vec
 Developed by a team at Google, Word2Vec aims to capture
the semantic relationships between words by mapping them
to high-dimensional vectors.
 The underlying idea is that words with similar meanings
should have similar vector representations.
 In Word2Vec every word is assigned a vector. We start with
either a random vector or one-hot vector
neural embedding methods
 Continuous Bag of Words (CBOW)
 Skip-gram.
Continuous Bag of Words(CBOW)
 Continuous Bag of Words (CBOW) is a type of neural
network architecture used in the Word2Vec model.
 The primary objective of CBOW is to predict a target word
based on its context, which consists of the surrounding
words in a given window.
 Given a sequence of words in a context window, the model is
trained to predict the target word at the center of the
window.
CBOW example
Architecture
CBOW
 The hidden layer contains the continuous vector
representations (word embeddings) of the input words.
 The weights between the input layer and the hidden layer are
learned during training.
 The dimensionality of the hidden layer represents the size of
the word embeddings (the continuous vector space).
The Skip-Gram model
 The Skip-Gram model learns distributed representations of
words in a continuous vector space.
 The main objective of Skip-Gram is to predict context words
(words surrounding a target word) given a target word.
 This is the opposite of the Continuous Bag of Words
(CBOW) model, where the objective is to predict the target
word based on its context.
 It is shown that this method produces more meaningful
embeddings.
Skipgram
Skipgram
 After applying the above neural embedding methods we get
trained vectors of each word after many iterations through
the corpus.
 These trained vectors preserve syntactical or semantic
information and are converted to lower dimensions.
 The vectors with similar meaning or semantic information
are placed close to each other in space.
Choice
 In practice, the choice between CBOW and Skip-gram often
depends on the specific characteristics of the data and the
task at hand.
 CBOW might be preferred when training resources are
limited, and capturing syntactic information is important.
 Skip-gram, on the other hand, might be chosen when
semantic relationships and the representation of rare words
are crucial.
Need for Word Embedding?
 To reduce dimensionality
 To use a word to predict the words around it.
 Inter-word semantics must be captured.
How are Word Embeddings used?
 They are used as input to machine learning models.
Take the words —-> Give their numeric representation —> Use in training or inference.
 To represent or visualize any underlying patterns of usage in
the corpus that was used to train them.
Discourse in NLP
 Discourse in NLP is nothing but coherent groups of sentences.
 When we are dealing with Natural Language Processing, the
provided language consists of structured, collective, and consistent
groups of sentences, which are termed discourse in NLP.
 The relationship between words makes the training of the NLP
model quite easy and more predictable than the actual results.
 Discourse Analysis is extracting the meaning out of the corpus
or text. Discourse Analysis is very important in Natural language
Processing and helps train the NLP model better.
Coherence
 Coherence in terms of Discourse in NLP means making
sense of the utterances or making meaningful connections
and correlations.
 There is a lot of connection between the coherence and the
discourse structure (discussed in the next section).
 We use the property of good text, coherence, etc., to
evaluate the quality of the output generated by the natural
language processing generation system.
Coherance
 What are coherent discourse texts?
 Well, if we read a paragraph from a newspaper, we can see
that the entire paragraph is interrelated; hence we can say
that the discourse is coherence, but if we only combine the
newspaper headlines consecutively, then it is not a discourse,
it is just a group of sentences that are also non-coherence.
Coherence Relation Between Utterances
 When we say that the discourses are coherent, then it simply
means that the discourse has some sort of meaningful
connection. ‘
 The coherent relation tells us that there is some sort of
connection present between the utterances.
 If there is some kind of relationship between the entities,
then we can also say that the discourse in NLP is coherent.
So, the coherence between the entities is known as entitybased coherence.
Discourse segmentation
 The segmentation is a difficult thing to implement, but it is
very necessary as discourse segmentation is used in fields like
 Information Retrieval,
 Text summarization,
 Information Extraction, etc.
Algorithms for Discourse Segmentation
 We have different algorithms for
 Unsupervised Discourse Segmentation
 Supervised Discourse Segmentation
Unsupervised Discourse Segmentation
 The class of unsupervised segmentation is also termed or
represented as linear segmentation
 Suppose we have a text with us, and the task is to segment the text
into various units of multi-paragraphs. In the multi-paragraphs, a
single unit is going to represent a passage of the text.
 unsupervised discourse segmentation means the classification and
grouping up of similar texts with the help of coherent discourse in
NLP.
 The unsupervised discourse segmentation can also be
performed with the help of lexicon cohesion. The lexicon
cohesion indicates the relationship among similar units, for
example, synonyms.
Supervised Discourse Segmentation
 In the previous segmentation, there was no certain labeled
segment boundary to separate the discourse segments.
 But in the supervised discourse segmentation, we only deal
with the training data set having a labeled boundary.
 To differentiate or structure the discourse segments, we
make use of cue words or discourse makers.
 These cue words or discourse maker works to signal the
discourse structure.
 As there can be varied domains of discourse in NLP so, the
cue words or discourse makers are domain specific.
Text Coherence
 Lexical repetition is a way to find the structure in a
discourse, but it does not satisfy the requirement of being
coherent discourse.
 To achieve the coherent discourse, we must focus on
coherence relations in specific.
 As we know that coherence relation defines the possible
connection between utterances in a discourse.
Text coherance
 We are taking two terms S0 and S1 to represent the meaning of the
two related sentences −
 Result
 It infers that the state asserted by term S0 could cause the state
asserted by S1. For example, two statements show the relationship
result: Ram was caught in the fire. His skin burned.
 Explanation
 It infers that the state asserted by S1 could cause the state asserted
by S0. For example, two statements show the relationship − Ram
fought with Shyam’s friend. He was drunk.
Parallel
 It infers p(a1,a2,…) from assertion of S0 and p(b1,b2,…)
from assertion S1.
 Here ai and bi are similar for all i.
 For example, two statements are parallel −
 Ram wanted car.
 Shyam wanted money.
Elaboration
 It infers the same proposition P from both the assertions
− S0 and S1
 For example, two statements show the relation elaboration:
 Ram was from Chandigarh.
 Shyam was from Kerala.
Occasion
 It happens when a change of state can be inferred from the
assertion of S0, final state of which can be inferred
from S1 and vice-versa.
 For example, the two statements show the relation occasion:
 Ram picked up the book.
 He gave it to Shyam.
Hierarchical structure
 S1 − Ram went to the bank to deposit money.
 S2 − He then took a train to Shyam’s cloth shop.
 S3 − He wanted to buy some clothes.
 S4 − He do not have new clothes for party.
 S5 − He also wanted to talk to Shyam regarding his health
Reference Resolution
 Interpretation of the sentences from any discourse is another
important task and to achieve this we need to know who or
what entity is being talked about.
 Here, interpretation reference is the key element.
 Reference may be defined as the linguistic expression to
denote an entity or individual. For example, in the
passage, Ram, the manager of ABC bank, saw his friend
Shyam at a shop.
 He went to meet him, the linguistic expressions like Ram,
His, He are reference.
Terminology Used in Reference Resolution
 Referring expression − The natural language expression
that is used to perform reference is called a referring
expression. For example, the passage used above is a
referring expression.
 Referent − It is the entity that is referred. For example, in
the last given example Ram is a referent.
 Corefer − When two expressions are used to refer to the
same entity, they are called corefers. For
example, Ram and he are corefers.
Terminology Used in Reference Resolution
 Antecedent − The term has the license to use another
term. For example, Ram is the antecedent of the
reference he.
 Anaphora & Anaphoric − It may be defined as the
reference to an entity that has been previously introduced
into the sentence. And, the referring expression is called
anaphoric.
 Discourse model − The model that contains the
representations of the entities that have been referred to in
the discourse and the relationship they are engaged in.
Types of Referring Expressions
 Indefinite Noun Phrases
 Definite Noun Phrases
 Pronouns
 Demonstratives
 Names
Indefinite Noun Phrases
 Such kind of reference represents the entities that are new to
the hearer into the discourse context.
 For example − in the sentence Ram had gone around one
day to bring him some food − some is an indefinite
reference.
Definite Noun Phrases
 Opposite to above, such kind of reference represents the
entities that are not new or identifiable to the hearer into the
discourse context.
 For example, in the sentence –
 I used to read The Times of India
 The Times of India is a definite reference.
Pronouns
 It is a form of definite reference. For example, Ram laughed
as loud as he could. The word he represents pronoun
referring expression.
Demonstratives
 These demonstrate and behave differently than simple
definite pronouns. For example, this and that are
demonstrative pronouns.
Names
 It is the simplest type of referring expression. It can be the
name of a person, organization and location also. For
example, in the above examples, Ram is the name-refereeing
expression.
Reference Resolution Tasks
Coreference Resolution
 It is the task of finding referring expressions in a text that
refer to the same entity.
 In simple words, it is the task of finding corefer expressions.
 A set of coreferring expressions are called coreference chain.
For example –
 He, Chief Manager and His - these are referring expressions
in the first passage given as example.
Pronominal Anaphora Resolution
 Unlike the coreference resolution, pronominal anaphora
resolution may be defined as the task of finding the
antecedent for a single pronoun.
 For example,
 the pronoun is his and the task of pronominal anaphora
resolution is to find the word Ram because Ram is the
antecedent.
Coreference Resolution
 Coreference resolution is the task of finding all referring
expressions like — (he, I, that, this…, or any subject or noun /
called mentions) is referred to which entity (referents like any
person, thing, subject etc...)
 After finding and grouping these mentions we can resolve them by
replacing, as stated above, pronouns with noun phrases.
 Example:
Coreference Resolution - Example
There are two entities in this sentence:
1.Naomi
2.Naomi's dress
• Naomi, her and she all refer to a single entity.
• Naomi's dress and it both refer to another entity.
Our brains will instantly recognize this, but a computer will not.
Coreference Resolution
 Coreference resolution is an exceptionally versatile tool and can
be applied to a variety of NLP tasks such as text understanding,
information extraction, machine translation, sentiment analysis, or
document summarization.
 It is a great way to obtain unambiguous sentences which can be
much more easily understood by computers.
Types of Coreference resolution
 Anaphora
 Cataphora
 Split Antecedents
Types of Coreference resolution
 Anaphora - “Anaphora is the use of an expression whose
interpretation depends specifically upon another (antecedent)
expression” or you can say “when the referring expression(anaphor)
is pointing backwards”
 Example:- The music was so loud that it couldn’t be enjoyed
 The word "it" is an example of an anaphor.
 It's a pronoun that refers back to the noun "music" mentioned
earlier in the sentence.
 So, "it" is used to avoid repeating the word "music" and still make
the sentence clear.
Types of Coreference resolution
 Cataphora - its said to be just reverse of anaphora →“the use of an
expression that depends upon a postcedent expression” or you can
say “when the referring expression is pointing forward”
 Example:- When he arrived home, John went to sleep.
 he is referred to John, and “he” came before “John” in sentence.
Types of Coreference resolution
 Split antecedents: It’s an anaphoric expression where the
pronoun (2) refers to more than one antecedent (1).
Some other examples of coreference resolution
 Example of named mentions instead of pronouns:-
1. International Business Machines sought patent compensation
from Amazon; IBM had previously sued other companies.
 Well you can see IBM referred to International Business
Machines…. these type of references are also there..
 2. Barack Obama traveled to …. Obama …
 So we can see “obama” and “Barack Obama” are referred to same
person.
What is coreference resolution?
 Coreference resolution thus comprises two tasks (although they are often
performed jointly):

(1) identifying the mentions
 (2) clustering them into coreference chains/discourse entities
What is coreference resolution?
Coreference resolution task
 The coreference resolution task is separated into two sub tasks:
1. Mention Detection
2. Mention Clustering
Coreference resolution task - Mention Detection
Mention Detection:
 In this sub-task the main goal is to find all the candidates spans
referring to some entities.
 For example, in the sentence below the mention detection step will
color all the candidates spans in blue:
• There are three kinds of mentions that will be detected in this
step: Pronoun, Named entity recognition, Noun phrases
1. Pronouns:
 A pronoun is a word that substitutes for noun phrase and usually
involves anaphora, where the meaning of the pronoun is dependent
on an antecedent.
 For instance, ‘She’ is the pronoun in the sentence “Noa gave an
amazing lecture, when she was in the conference at Madrid last
year.”
Coreference resolution task - Mention Detection
Coreference resolution task - Mention Detection
2. Named-entity recognition (NER):
NER model locates entities in unstructured text and classify them into pre-defined
categories (such as person names, organizations, locations, products, etc).
Coreference resolution task - Mention Detection
3. Noun-phrases:
A noun phrase is a bunch words that is headed by a noun and includes modifiers
(e.g., ‘the,’ ‘a,’ ‘of them,’ ‘with her’).
Coreference resolution task - Mention Clustering
 Once we hold the mentions, the goal of the second sub-task is attempting to
identify which ones refer to the same entity. Then, merging the mentions into the
cluster corresponding to the entities presented in the text.
 Antecedent, Anaphora, Cataphora are the basic terms of mention
clustering.
Coreference resolution Models – Rule based models
Types of coreference resolution models
 1. Rule-Based Models
 2. Mention-Pair Models
 3. Mention-Ranking Models
Coreference resolution Models - Mention-Pair Models
 The second kind of coreference model (Mention-Pair Models) is a binary
classifier.
 The idea is to train a classifier that will find the probability of two mentions to
be coreferent.

It will compare between each two mentions in the sentence and will assign their
probability to be connected. The ideal condition will be that negative examples
will receive a probability that is closed to 0 and positive examples will be close
to 1.
Coreference resolution Models - Mention-Pair Models
Coreference resolution Models - Mention-Pair Models
Coreference resolution Models - Mention-Ranking Models
 The mention-ranking model assigns a score to each pair of candidate antecedent
and corefernts. In the end of the procedure the model will choose just one of the
pairs
Coreference resolution Models - Mention-Ranking Models
Coreference resolution Models - Mention-Ranking Models
https://galhever.medium.com/a-review-to-coreference-resolution-models-f44b4360a00
Coreference Resolution vs Anaphora Resolution
 Coreference Resolution (CR):

This is like connecting the dots between words or phrases that talk about the same thing.
Imagine you're reading a story, and the story mentions "Mary" and later says "She did
something." CR helps us figure out that "She" is talking about "Mary."
 It's like finding out that they're both talking about the same person.
 Anaphora Resolution (AR):
 This is a specific type of CR.
 It's like solving a little puzzle in a story. Anaphora means words like "he," "she," "it," or
"they" that refer to something already mentioned. For example, if a story says, "John found
a wallet. He returned it," AR helps us know that "He" is talking about "John" and "It" is
talking about the "wallet."
 So, CR is like connecting any two things that are talking about the same stuff, while AR is like
solving the puzzle of what words like "he" and "it" are talking about.
Coreference Resolution vs Anaphora Resolution
Scope aspect of
Coreference Resolution and Anaphora Resolution
 Coreference Resolution:
 Broader Scope:
 Definition: Coreference resolution identifies all instances in a text where different
expressions refer to the same entity.
 Types of References:
 Pronouns: Identifies when pronouns refer to the same entity (e.g., "Jane said she would go."
Here, "Jane" and "she" refer to the same person).
 Noun Phrases: Identifies when different noun phrases refer to the same entity (e.g., "The
president gave a speech. The leader of the nation was eloquent." Here, "The president" and
"The leader of the nation" refer to the same person).
 Named Entities: Identifies when different mentions of named entities refer to the same thing
(e.g., "IBM released a new product. The company has high hopes for it." Here, "IBM" and "The
company" refer to the same entity).
 Definite Descriptions: Identifies when definite descriptions refer to the same entity (e.g., "I
saw a dog.The dog was barking." Here, "a dog" and "The dog" refer to the same dog).
Scope aspect of
Coreference Resolution and Anaphora Resolution
 Anaphora Resolution:
 Narrower Scope:
 Definition: Anaphora resolution specifically deals with resolving references
where a word or phrase refers back to another word or phrase used earlier in
the text.
 Focus:
 Pronouns: Primarily focuses on resolving pronouns (e.g., "The cat sat on the mat. It
was sleepy." Here, "It" refers back to "The cat").
 Demonstratives: Resolves demonstratives like "this," "that," "these," and "those"
when they refer back to something mentioned earlier (e.g., "I like apples. These are
my favorite fruit." Here, "These" refers back to "apples").
 Other Similar References: Can also resolve other types of references like "such"
or "same" when they refer back to something earlier in the text (e.g., "He gave a
speech. Such eloquence is rare." Here, "Such" refers back to "a speech").