Building a Dataset for Possessions Identification in Text

Building a Dataset for Possessions Identification in Text Carmen Banea, Xi Chen, Rada Mihalcea Electrical Engineering and Computer Science University of Michigan Ann Arbor, Michigan carmennb@umich.edu, chenxicx@umich.edu, mihalcea@umich.edu Abstract Just as industrialization matured from mass production to customization and personalization, so has the Web migrated from generic content to public disclosures of one’s most intimately held thoughts, opinions and beliefs. This relatively new type of data is able to represent finer and more narrowly defined demographic slices. If until now researchers have primarily focused on leveraging personalized content to identify latent information such as gender, nationality, location, or age of the author, this study seeks to establish a structured way of extracting possessions, or items that people own or are entitled to, as a way to ultimately provide insights into people’s behaviors and characteristics. In order to promote more research in this area, we are releasing a set of 798 possessions extracted from blog genre, where possessions are marked at different confidence levels, as well as a detailed set of guidelines to help in future annotation studies. Keywords: possession identification, latent attribute extraction, author profiling 1. Introduction “Watch your thoughts, they become your words. Watch your words, they become your actions. Watch your actions, they become your habits. Watch your habits, they become your character. Watch your character, it becomes your destiny.” Author Unknown With the introduction and adoption of Web 2.0, the In- ternet has become a forum where users voice their opin- ions and feelings through comments, reviews, blogs, mi- croblogs, status updates, and other forms of online partic- ipation. This growing and diverse unstructured stream of information blends for the first time consumer demograph- ics, with lifestyle information and choices, user opinions and feelings, as well as mentions of items that are owned or liked by the user. Our research is motivated by the affective-cognitive consis- tency model (Rosenberg, 1956; Rosenberg, 1968), a branch of cognitive consistency theory that not only hypothesizes that people are motivated to seek a coherent state both in- ternally (at the level of thoughts, beliefs, feelings, and val- ues) and externally (through attitudes and behaviors), but also that individuals gain more motivation in achieving a consistent state so that others perceive them to be consis- tent. This particular model implies that in a public setting, where others are reading and scrutinizing a person’s online content, the blended data forms not only a raw user profile, but it encodes a coherent user signature, which, if mined successfully, can carve out very narrowly defined clusters of individuals who live, think and act alike. Motivated by the potential of analyzing and evaluating peo- ple’s behaviors and characteristics as they relate to their ownership of a given object, this work seeks to establish the framework for identifying such objects. To this end, we propose the concept of “possessions,” or textual represen- tation of items that somebody owns or is entitled to, which we define in detail in Section 2.1. To date, research focusing on extracting latent …

Content truncated to 500 words