SEMANTIC ANALYSIS

Semantic Analysis

•Definition: Understanding the meaning of words and phrases in a text after parsing
its syntax.
•Target language: A representation language, including the framework and
vocabulary, depends on the purpose of the natural language understanding system.
•Two types of target languages:
•One designed for language processing with primitives that correspond to language
distinctions.
•One aligned with the requirements of a backend program.
•Role of semantic processing: Creates a target language representation and imposes
constraints based on world knowledge and helps select among competing syntactic
analyses.
•Example: "Is the glass jar peanut butter?" - Semantic processing can impose
constraints to determine if the sentence makes sense based on world knowledge.
Lexical Processing

• Lexical processing is the first step in any semantic processing system

• The goal is to extract the meanings of individual words from a dictionary
or lexicon
• Many words have multiple meanings, which can lead to lexical ambiguity
• Word sense disambiguation is the process of determining the correct
meaning of a word based on its context
• Semantic markers, such as PHYSICAL-OBJECT, ANIMATE-OBJECT, and
ABSTRACT-OBJECT, can be used to help disambiguate word senses
Sentence-Level Processing

• Semantic grammars combine syntactic, semantic, and pragmatic knowledge to

create a description of a sentence. Parsing with this type of grammar results in
a semantic representation that goes beyond a mere syntactic description.
• Case grammars build a structure with some semantic information, but further
interpretation may be necessary.
• Conceptual parsing combines syntactic and semantic knowledge into a single
interpretive system that prioritizes semantic interpretation over syntax.
• Approximately compositional semantic interpretation applies semantic
processing to the results of a syntactic parse, either incrementally or all at once
for complete sentences.
Semantic Grammars

• Semantic grammars combine syntactic, semantic, and pragmatic

knowledge into a set of rules in the form of a grammar.
• The result of parsing with a semantic grammar is a more meaningful and
complete description of a sentence than just a syntactic description.
• Semantic grammars help resolve lexical ambiguity by using context and
word sense disambiguation.
• They also consider the pragmatic context and the intended meaning of
the sentence.
Semantic Grammars Example

• Consider the sentence: "John is taller than Mary."

• A semantic grammar for this sentence might have the following rules:
• John and Mary are people.
• Taller is a comparative adjective.
• The comparative form of tall takes two arguments: the thing being compared (John) and the thing being
compared to (Mary).
• The result of the comparison is a truth value (i.e., true or false).
• Using these rules, the semantic representation of the sentence could be generated as
follows:
• Comparison(John, Mary, Taller) = True
• used for various natural language processing tasks, such as answering questions like "Who is
taller, John or Mary?" or "Is Mary taller than John?"
 (S)
|
-------------------
| |
(NP) (VP)
| |
----- -------------
| | | |
John (BE Verb) (AdjP) (PP)
| | | |
taller than Mary
Case Grammars

• Case Grammars are a type of approach used for creating a semantic

representation of a sentence. In this approach, the structure built by the
parser contains some semantics, although further interpretation may also
be necessary.
• The idea behind case grammars is to identify and label the core elements
of a sentence, such as the subject, object, and verb, and then use this
information to generate a semantic representation. The core elements are
labeled using case markers, which indicate their semantic role in the
sentence.
Example

Consider the sentence "The cat chased the mouse." A Case Grammar for
this sentence would be structured as follows:
• The sentence consists of a subject noun phrase and a predicate verb
phrase.
• The subject noun phrase consists of the determiner "the" and the noun
"cat."
• The predicate verb phrase consists of the verb "chased" and the object
noun phrase.
• The object noun phrase consists of the determiner "the" and the noun
"mouse."
 Sentence
/\
/ \
/ \
NP VP
/\ /\
/ \ / \
Det N V NP
| | | /\
the cat chased Det N
| |
the mouse
Conceptual Parsing

• Conceptual parsing is an approach to sentence-level processing that combines

both syntactic and semantic knowledge into a single system.
• Syntactic parsing is used to create a structural representation of a sentence,
while semantic interpretation is used to drive the understanding of the meaning
of the sentence.
• This approach is useful because it allows the system to understand the meaning
of a sentence as a whole, rather than relying on individual words or phrases.
• Conceptual parsing is often used in natural language processing applications,
such as chatbots and virtual assistants, to improve their ability to understand
and respond to user inputs.