David B. Bracewell

A Motif Approach for Identifying Pursuits of Power in Social Discourse

In this paper, we investigate whether the social goals of an individuals utterances can be recognized through analysis of a discourses intentional structure. Specifically we focus on identifying individuals pursuing power within a group. Individuals pursue power in order to increase their control of the actions and goals of the group. Following work in discourse processing we decompose the problem into identifying the social intention of the discourse segments and the intentional structure of the overall discourse. The set of social intentions come from eight psychologically-motivated social acts. We then build a motif-based representation of the discourses social intentional structure that captures the interactions of the intentions between discourse participants. Finally we show how these structures can be used to identify the social goal of pursuit of power. Our best results achieve an accuracy of 84.2% for predicting pursuit of power in discussions communicated in English and 80.6% for discussions communicated in Chinese.

Determining the conceptual space of metaphoric expressions

We present a method of constructing the semantic signatures of target concepts expressed in metaphoric expressions as well as a method to determine the conceptual space of a metaphor using the constructed semantic signatures and a semantic expansion. We evaluate our methodology by focusing on metaphors where the target concept is Governance. Using the semantic signature constructed for this concept, we show that the conceptual spaces generated by our method are judged to be highly acceptable by humans.

Who's Playing Well with Others: Determining Collegiality in Text

In this paper, we present a framework for determining the interpersonal relations exhibited between two individuals. Specifically, we focus on recognizing the presence or absence of collegiality in discussion threads and dialogues. Collegiality results from the existence of harmonious relationships irrespective of the groups power structure. We have identified four psychologically-motived language uses that indicate collegiality. These language uses are identified in text with the use of a set of attributes that are assigned to each language use and can be extracted using grammars and lexicons. Through the attributes, language uses, and dialogue features, a model can be learned that can determine whether two people are collegial, uncollegial, or whether there is not enough information. Using multi-class logistic regression, we obtain an overall micro-averaged F-measure of 83.3%.

Applying Textual Entailment to the Interpretation of Metaphor

Metaphor is a pervasive feature of human language that enables us to conceptualize and communicate abstract concepts using more concrete terminology. Unfortunately, computational models of natural language understanding - including systems for question answering, textual entailment, lexical substitution, and word-sense disambiguation - are unable to appropriately grasp the semantic content of metaphor and other forms of figurative language. In particular, we address the problem of understanding metaphoric language in the context of entailment (or paraphrase) detection. We build upon our existing state-of-the-art textual entailment system to specifically address issues of lexical entailment within a metaphoric context and have performed an in-depth experimental analysis to determine which techniques are most effective at interpreting metaphorical text. Our results suggest that a machine learning system trained on metaphor-rich data can achieve an accuracy above 90% for verbal metaphors using a combination of lexical, semantic, and contextual measures of term similarity.

The Needs of Metaphor

In this paper we semi-automatically construct a multilingual lexicon for Maslow’s seven categories of needs. We then use the semi-automatically constructed lexicons and a metaphor recognition system to analyze the change in rate of the expression of needs in the presence of metaphor. We examine four languages, English, Farsi, Russian, and Spanish, and focus on metaphors whose target concept is related to poverty or taxation.

#impressme: The Language of Motivation in User Generated Content

An individuals ability to produce quality work is a function of their current motivation, their control over the results of their work, and the social influences of other individuals. All of these factors can be identified in the language that individuals use to discuss their work with their peers. Previous approaches to modeling motivation have relied on social-network and time-series analysis to predict the popularity of a contribution to user-generated content site. In contrast, we show how an individuals use of language can reflect their level of motivation and can be used to predict their future performance. We compare our results to an analysis of motivation based on utility theory. We show that an understanding of the language contained in comments on user generated content sites provides significant insight into an authors level of motivation and the potential quality of their future work.

In you we follow: determining the group leader in dialogue

In this paper, we investigate whether the social roles of dialogue participants can be recognized through the social actions performed by the participant in their interactions with others in the group. Specifically we focus on determining if a participant is the leader of the group. We decompose the problem into identifying the social goals for participant discourse segments. These social goals are represented through a set of eleven psychologically-motivated social acts. We then model leadership using a sociological-inspired model called social rank which takes into account the social capital accumulated by the participant over the course of a single dialogue. We explore these models in task-oriented dialogues communicated in English, Arabic, and Chinese and show that the incorporation of social rank can improve precision of detecting the leader by 14% in English, 8% in Arabic, and 4% in Chinese.

Semi-supervised Modeling of Social Actions in Online Dialogue

The emergence of discussion and debate on social media necessitates the development of new models for processing dialogue. Vitally important to inferring the social implicatures of dialogue on social media is to understand the social goals and desires of the participants. Thus, to infer social implicatures methods for capturing the social goals and intentions of the participants must be first developed. In this paper, we propose a set of fifteen social acts to infer the social goals of dialogue participants. Social acts capture the complex social actions individuals signal through their utterances. We present a semi-supervised algorithm called the Social Act Conversation Model (SACM) for the fifteen social acts. The algorithm is based on the premise that linguistic expressions in social dialogue relate directly to the topic being discussed or to the social actions of the participants. We show that incorporating the social acts identified by the SACM with existing pattern-based identification can increase the performance in inferring social implicatures (adversarial behavior, pursuing power, and leadership) for online dialogue communicated in English, Arabic, and Chinese.

The Social Actions of Successful Groups

The success of a group is determined by a number of factors. Some of these factors, such as task difficulty and availability of resources, are out of a groups control and are constant amongst competing teams working on the same task. One factor which is not constant is the social dynamics of the group. The mix of behaviors by individuals in the group, as seen through social relations, social actions, and social roles are key in determining a groups success. In this paper, we examine if it is possible to determine the success, or status, of a Wikipedia article through the social dynamics of the associated discussions. We capture social dynamics using four higher level social phenomena and fourteen lower level social acts, which we define from prevailing theories of group success. We examine discussions around English and Chinese Wikipedia articles and find that using social acts can increase prediction over standard network metric approaches by 35.0% for English to 90.3% and 12.3% for Chinese to 88.7%.