Cecilia Ovesdotter Alm

Emotions from Text: Machine Learning for Text-based Emotion Prediction

In addition to information, text contains attitudinal, and more specifically, emotional content. This paper explores the text-based emotion prediction problem empirically, using supervised machine learning with the SNoW learning architecture. The goal is to classify the emotional affinity of sentences in the narrative domain of childrens fairy tales, for subsequent usage in appropriate expressive rendering of text-to-speech synthesis. Initial experiments on a preliminary data set of 22 fairy tales show encouraging results over a naive baseline and BOW approach for classification of emotional versus non-emotional contents, with some dependency on parameter tuning. We also discuss results for a tripartite model which covers emotional valence, as well as feature set alternations. In addition, we present plans for a more cognitively sound sequential model, taking into consideration a larger set of basic emotions.

English in the Ecuadorian commercial context

This paper presents a study completed in Quito, Ecuadors capital, in 2002. It investigates the attitudinal perceptions toward English in advertising in this context, as well as the actual distribution of English in magazine ads and commercial names of business establishments. The findings are the result of four data collection procedures: first, a questionnaire administered to advertising experts; second, an analysis of business names in ten shopping centers; third, an analysis of advertisements in Ecuadorian magazines; and fourth, an interview survey with the same group of advertising experts. The results are analyzed both quantitatively and qualitatively, with the aim to provide an attitudinal sociolinguistic profile of English in Ecuador from a descriptive, comparative and critical perspective. Adopting the socioeconomic framework presented by Bourdieu (1991), English is found to represent commercial capital. Moreover, English is shown to be highly stratified according to socioeconomic strata, and to function as a segmentizer and a gatekeeper on the Ecuadorian market. Thus, if English is to succeed in functioning as empowerment (cf. Friedrich, 2001) among the disadvantaged in Ecuador in the future, affirmative action is needed, especially within the educational sector.

Discriminating Image Senses by Clustering with Multimodal Features

We discuss Image Sense Discrimination (ISD), and apply a method based on spectral clustering, using multimodal features from the image and text of the embedding web page. We evaluate our method on a new data set of annotated web images, retrieved with ambiguous query terms. Experiments investigate different levels of sense granularity, as well as the impact of text and image features, and global versus local text features.

Emotional sequencing and development in fairy tales

Affect is a transient phenomenon, with emotions tending to blend and interact over time [4]. This paper discusses emotional distributions in child-directed texts. It provides statistical evidence for the relevance of emotional sequencing, and evaluates trends of emotional story development, based on annotation statistics on 22 Grimms’ fairy tales which form part of a larger on-going text-annotation project that is also introduced. The study is motivated by the need for exploring features for text-based emotion prediction at the sentence-level, for use in expressive text-to-speech synthesis of children’s stories.

Toward Macro-Insights for Suicide Prevention: Analyzing Fine-Grained Distress at Scale

Suicide is a leading cause of death in the United States. One of the major challenges to suicide prevention is that those who may be most at risk cannot be relied upon to report their conditions to clinicians. This paper takes an initial step toward the automatic detection of suicidal risk factors through social media activity, with no reliance on self-reporting. We consider the performance of annotators with various degrees of expertise in suicide prevention at annotating microblog data for the purpose of training text-based models for detecting suicide risk behaviors. Consistent with crowdsourcing literature, we found that novice-novice annotator pairs underperform expert annotators and outperform automatic lexical analysis tools, such as Linguistic Inquiry and Word Count.

An Analysis of Domestic Abuse Discourse on Reddit

Domestic abuse affects people of every race, class, age, and nation. There is significant research on the prevalence and effects of domestic abuse; however, such research typically involves population-based surveys that have high financial costs. This work provides a qualitative analysis of domestic abuse using data collected from the social and news-aggregation website reddit.com. We develop classifiers to detect submissions discussing domestic abuse, achieving accuracies of up to 92%, a substantial error reduction over its baseline. Analysis of the top features used in detecting abuse discourse provides insight into the dynamics of abusive relationships.

#WhyIStayed, #WhyILeft: Microblogging to Make Sense of Domestic Abuse

In September 2014, Twitter users unequivocally reacted to the Ray Rice assault scandal by unleashing personal stories of domestic abuse via the hashtags #WhyIStayed or #WhyILeft. We explore at a macro-level firsthand accounts of domestic abuse from a substantial, balanced corpus of tweeted instances designated with these tags. To seek insights into the reasons victims give for staying in vs. leaving abusive relationships, we analyze the corpus using linguistically motivated methods. We also report on an annotation study for corpus assessment. We perform classification, contributing a classifier that discriminates between the two hashtags exceptionally well at 82% accuracy with a substantial error reduction over its baseline.

Challenges for Annotating Images for Sense Disambiguation

We describe an unusual data set of thousands of annotated images with interesting sense phenomena. Natural language image sense annotation involves increased semantic complexities compared to disambiguating word senses when annotating text. These issues are discussed and illustrated, including the distinction between word senses and iconographic senses.

The Role of Affect in the Computational Modeling of Natural Language

Expressivity is an intrinsic component of natural language. This article follows the tradition of affective computing (Picard 1997) in using affect to refer to connected concepts such as emotion, mood, feelings, personality, attitude, polarity, and related subjective phenomena. The overview clarifies the relevance of affect for linguistics and computational linguistics, summarizes useful background, outlines helpful resources, and highlights important considerations for computational modeling of affect in language and affect-related linguistic behaviors. In addition, the article sketches unsettled debates, topics, problems, and areas in need of further exploration.

An Expert-in-the-loop Paradigm for Learning Medical Image Grouping

Image grouping in knowledge-rich domains is challenging, since domain knowledge and expertise are key to transform image pixels into meaningful content. Manually marking and annotating images is not only labor-intensive but also ineffective. Furthermore, most traditional machine learning approaches cannot bridge this gap for the absence of experts’ input. We thus present an interactive machine learning paradigm that allows experts to become an integral part of the learning process. This paradigm is designed for automatically computing and quantifying interpretable grouping of dermatological images. In this way, the computational evolution of an image grouping model, its visualization, and expert interactions form a loop to improve image grouping. In our paradigm, dermatologists encode their domain knowledge about the medical images by grouping a small subset of images via a carefully designed interface. Our learning algorithm automatically incorporates these manually specified connections as constraints for re-organizing the whole image dataset. Performance evaluation shows that this paradigm effectively improves image grouping based on expert knowledge.