Available online at www.sciencedirect.

com

Procedia - Social and Behavioral Sciences 00 (2013) 000000

6th International Conference EDU-WORLD 2014 - Education Facing Contemporary World

Issues

Analysis of Bridge Anaphora across novel

Daniela Gfu1, Alexandru Iliescu
Alexandru Ioan Cuza University of Iai, Romania

Abstract

The paper describes importance of anaphoric relations as semantic relations in Quo Vadis novel, in order to put in
evidence characters paths, in the benefit of readers (teachers, students, etc.). In fact, this research could help the
potential reader to understand in a short period of time if the novel responds or not his needs, according to what
he are looking for. For instance, if he/her prefers a particular author who has tackled several topics among his
writings, but not all of them would keep the passion of reading. Our method could help, in this case, to choose the
right book. Moreover, this paper presents a set of conventions of annotations for 11 bridge anaphora relations
(coref-interpret, member-of, has-as-member, isa, class-of, part-of, has-as-part, subgroup-of,
has-as-subgroup, has-name, name-of) as a result of manually annotations made by a team of trained
students in Computational Linguistics, being essential in this kind of study. Bridge anaphora relations, in this
context, include linguistic expressions about characters novel. In other words, we have different and complete
profiles of subjects necessary to understand their stories, including temporal and event information. Our
documentation reveals that the analysis of semantic relations was focused on corpora structured of online
newspapers, blogs, Wikipedia texts, etc. Here it is a space focused on a large diversity of authors, styles, issues,
unsupervised information, etc. The most important reason for this kind of analysis was choosing the fiction, as a
strategic structure followed by one author with a recognized style. Starting with manually annotations we
describe patterns which we intend to incorporate for automatically identification of bridge anaphora relations. An
accurate characters profile is dependent on a multitude of Natural Language Processing techniques. In this study
we tested Weka to preprocess raw trajectories in order to transform them into semantic trajectories. From this
point, we compared manually and automatic annotations to improve our work. The method can be extended to
any kind of texts, who intended to help linguistics, teachers, students, readers in general, which are motivated to
understand the entities profiles, in order to implement a tool for recognition of bridge anaphora relations.
Keywords: civic identity, pragmalinguistics analysis, semantic classes, media online, journalists.

* Corresponding author. Tel.: +40-742050673

E-mail address: daniela.gifu@info.uaic.ro, alexandruiliescu19@gmail.com