Pronoun Resolution

Jurafsky and Martin [JM] suggest an algorithm by Lappine and Leass (1994) for pronoun interpretation which takes many concerns into consideration. There are two types of operations that this algorithm performs:

Discourse model update
When a noun phrase evokes a new entity, a representation for it is added to the model and a degree of salience for it is computed. [Actually a vector of salience factors is created. The salience value is a weighted sum of the component of this vector.]

As each sentence is processed, each factor of the computed value is reduced by half.

Sentence recency 100
Subject emphasis 80
Existential emphasis 70
Accusative (direct object) emphasis 50
Indirect object and oblique complement emphasis 40
Non-adverbial emphasis 50
Head noun emphasis 80
Figure 18.5 Salience Factors in Lappin and Leass's system. [JM]

Pronoun resolution
Any pronoun encountered is then processed using the mechanisms presented above to determine the referrent with the highest salience.

In practice, each referent is associated with an equivalence class of all expressions which reference it. The referent's salience is then the sum of the max of each factor of elements within the equivalence class.

Examples

18.62 An Acura Integra is parked in the lot. (subject)
18.63 There is an Acura Integra parked in the lot. (existential predicate nominal)
18.64 John parked an Acura Integra in the lot. (object)
18.65 John gave his Acura Integra a bath. (indirect object)
18.66 Inside his Acura Integra, John showed Susan his new CD player. (demarcated adverbial prepositional phrase)