NLP Exam 2
Lexical Semantics - answer Focuses on the meaning of words and their relationships
What is similarity? - answer Is-A relationships between words
Ex: Apple is-a fruit
Cherry is-a fruit
Mango is-a fruit
What is relatedness? - answer Any relationships between words
Ex: Smoking and Heart Disease
Needle and Thread
Sugar and Diabetes
Why would we use Similarity and Relatedness? - answer For Information retrieval,
especially for query expansion in order to retrieve documents whose words have similar
meanings to the query words
Word Sense Disambiguation
Similarity Measures - answerPath-based
Information-Content Based
Path-based Measures - answerUse the path information between concepts in order to
quantify their similarity
Information Content Measures - answerUtilize path information but also incorporate the
probability of the concept occurring in text
Relatedness Measures - answerUtilize some contextual information about the concepts
Definition based
Distributional based
Definition-Based Measures - answerAdapted Lesk: based on overlaps between the two
concept's extended definitions
∑(overlap length) ^ 2
Distributional Measures - answerNumerically calculate the similarity between terms
Co-occurrence Vectors - answerMeasure frequencies/probabilities/existence of words
occurring in the definition
Cosine Similarity - answerCosine(X Y) = (X·Y)/(|X||Y|)
Lexical Semantics - answer Focuses on the meaning of words and their relationships
What is similarity? - answer Is-A relationships between words
Ex: Apple is-a fruit
Cherry is-a fruit
Mango is-a fruit
What is relatedness? - answer Any relationships between words
Ex: Smoking and Heart Disease
Needle and Thread
Sugar and Diabetes
Why would we use Similarity and Relatedness? - answer For Information retrieval,
especially for query expansion in order to retrieve documents whose words have similar
meanings to the query words
Word Sense Disambiguation
Similarity Measures - answerPath-based
Information-Content Based
Path-based Measures - answerUse the path information between concepts in order to
quantify their similarity
Information Content Measures - answerUtilize path information but also incorporate the
probability of the concept occurring in text
Relatedness Measures - answerUtilize some contextual information about the concepts
Definition based
Distributional based
Definition-Based Measures - answerAdapted Lesk: based on overlaps between the two
concept's extended definitions
∑(overlap length) ^ 2
Distributional Measures - answerNumerically calculate the similarity between terms
Co-occurrence Vectors - answerMeasure frequencies/probabilities/existence of words
occurring in the definition
Cosine Similarity - answerCosine(X Y) = (X·Y)/(|X||Y|)
