Connotation Frames: A Data-Driven Investigation

Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, pages 311–321, Berlin, Germany, August 7-12, 2016. c⃝2016 Association for Computational Linguistics Connotation Frames: A Data-Driven Investigation Hannah Rashkin Sameer Singh Computer Science & EngineeringYejin Choi University of Washington {hrashkin, sameer, yejin}@cs.washington.edu Abstract Through a particular choice of a predicate (e.g., “x violated y”), a writer can subtly connote a range of implied sentiment and presupposed facts about the entities x and y: (1) writer’s perspective: projecting x as an “antagonist” and y as a “victim”, (2) entities’ perspective: y probably dislikes x, (3) effect: something bad happened to y, (4) value: y is something valuable, and (5) mental state: y is distressed by the event. We introduce connotation frames as a rep- resentation formalism to organize these rich dimensions of connotation using typed relations. First, we investigate the fea- sibility of obtaining connotative labels through crowdsourcing experiments. We then present models for predicting the con- notation frames of verb predicates based on their distributional word representations and the interplay between different types of connotative relations. Empirical results confirm that connotation frames can be in- duced from various data sources that reflect how language is used in context. We con- clude with analytical results that show the potential use of connotation frames for an- alyzing subtle biases in online news media. 1 Introduction People commonly express their opinions through subtle and nuanced language (Thomas et al., 2006; Somasundaran and Wiebe, 2010). Often, through seemingly objective statements, the writer can in- fluence the readers’ judgments toward an event and their participants. Even by choosing a particular predicate, the writer can indicate rich connotative information about the entities that interact through the predicate. More specifically, through a simple Writer: “Agent violates theme.” Writer + - - - - = + - Reader - = = Agent Theme P(w ! agent) P(w ! theme) P(agent ! theme) E(agent) E(theme) V(theme) V(agent) E(theme) S(agent) S(agent) E(agent) Perspective: the writer is sympathetic towards the theme Perspective: the writer portrays the agent as being antagonistic Value: the theme must be valuable + Effect: the agent is not really affected by the violation State: the theme will be unhappy State: the agent feels indifferent Effect: the theme has been hurt Value: not clear if agent is valuable Figure 1: An example connotation frame of “violate” as a set of typed relations: perspective P(x → y), effect E(x), value V(x), and mental state S(x). statement such as “x violated y”, the writer can convey: (1) writer’s perspective: the writer is projecting x as an “antagonist” and y as a “victim”, elic- iting negative perspective from readers toward x (i.e., blaming x) and positive perspective to- ward y (i.e., sympathetic or supportive toward y). (2) entities’ perspective: y most likely feels neg- atively toward x as a result of being violated. (3) effect: something bad happened to y. (4) value: y is something valuable, since it does not make sense to violate something worthless. In other words, …

Content truncated to 500 words