engleski

Inferring standard name form, gender and nobility from historical texts using stable model semantics

In this paper, we attack the problem of parsing name expressions and inferring standard name form, gender and nobility status from serial historical sources. This is a small but important part of modelling historians’ analysis of such sources, as they extract a lot of information from the names in text, and this information constrain their search. The task of parsing proper names seems to be easy, but it is a hard problem even for the modern languages, and even more challenging for the languages of historical sources. The test case used for the research was from the middle 19th century census for the old town centre of Zagreb. In order to evaluate and compare the fitness of the probabilistic and rule-based models for the task of inferring standard name form, both conditional random field (CRF) and rule-based models based on stable model semantics (Answer Set Programming Rules) were developed. Our results indicated that the rule- based approach is more suitable for inferring standard name forms from historical texts than the more widespread statistical approach.

parsing name ; historical texts ; semantics logic

nije evidentirano