You have seen that concepts and terms are closely related to each other- terms acting as handles for concepts .Let`s now look at two other concepts which will helps us refine our representation of meaning further- entities and entity types.
To summarise, you learnt that entities are instances of entity types. Also you saw that multiple entity types can be grouped under a concept.
You will see that we can use this hierarchical model of concepts, entities and entity types to answer some real-world questions and process the ‘meaning’ of sentences. cept.
You will see that we can use this hierarchical model of concepts, entities and entity types to answer some real-world questions and process the ‘meaning’ of sentences.
Associations Between Entity Types
Say you ask Alexa ,Alexa ,wha t is a Labrador?,it is a breed of dogs.Say you ask `Who is the coah of Indian crickt team?, it reveals the name of the coach.
To answer such questions, a system needs some kind of mapping between entities and entity types . i.e.. it needs to understand that a Labrador is a dog, a mammal is an animal, a coach is a specific person etc.
This brings us to the concept of associations between entities and entity types. These associations are represented useing the notion of predicates.Let’s study the idea of predicates.
To summarise, you saw that the notion of a predicate gives us a simple model to process the meaning of complex statements.For example ,say you ask an artificial NLP system – “Did France win the football world cup final in 2018?”. The statement can be broken down into a set of predicates, each returning True or False, such as win(France, final) = True, final(FIFA, 2018) = True.
A predicate is a function which takes in some parameters and returns True or False depending on the relationship between the parameters. For example, a predicate teacher_teaches_course(P = professor Srinath, C = text analytics) returns True.
You also saw that processing meaning involves understanding the associations between entities. For e.g. “A football team is comprised of 11 players” is an example of relation type =2 (since there are two entity types – team & players), though the relationship instance is 11 (since there are 11 instances of the entity).