Hemant Purohit

What kind of #conversation is Twitter? Mining #psycholinguistic cues for emergency coordination

The information overload created by social media messages in emergency situations challenges response organizations to find targeted content and users. We aim to select useful messages by detecting the presence of conversation as an indicator of coordinated citizen action. Using simple linguistic indicators drawn from conversation analysis in social science, we model the presence of coordination in the communication landscape of Twitter using a corpus of 1.5 million tweets for various disaster and non-disaster events spanning different periods, lengths of time, and varied social significance. Within replies, retweets and tweets that mention other Twitter users, we found that domain-independent, linguistic cues distinguish likely conversation from non-conversation in this online form of mediated communication. We demonstrate that these likely conversation subsets potentially contain more information than non-conversation subsets, whether or not the tweets are replies, retweets, or mention other Twitter users, as long as they reflect conversational properties. From a practical perspective, we have developed a model for trimming the candidate tweet corpus to identify a much smaller subset of data for submission to deeper, domain-dependent semantic analyses for the identification of actionable information nuggets for coordinated emergency response.

Intent Classification of Short-Text on Social Media

Social media platforms facilitate the emergence of citizen communities that discuss real-world events. Their content reflects a variety of intent ranging from social good (e.g., volunteering to help) to commercial interest (e.g., criticizing product features). Hence, mining intent from social data can aid in filtering social media to support organizations, such as an emergency management unit for resource planning. However, effective intent mining is inherently challenging due to ambiguity in interpretation, and sparsity of relevant behaviors in social data. In this paper, we address the problem of multiclass classification of intent with a use-case of social data generated during crisis events. Our novel method exploits a hybrid feature representation created by combining top-down processing using knowledge-guided patterns with bottom-up processing using a bag-of-tokens model. We employ pattern-set creation from a variety of knowledge sources including psycholinguistics to tackle the ambiguity challenge, social behavior about conversations to enrich context, and contrast patterns to tackle the sparsity challenge. Our results show a significant absolute gain up to 7% in the F1 score relative to a baseline using bottom-up processing alone, within the popular multiclass frameworks of One-vs-One and One-vs-All. Intent mining can help design efficient cooperative information systems between citizens and organizations for serving organizational information needs.

Mining Help Intent on Twitter During Disasters via Transfer Learning with Sparse Coding.

Citizens share a variety of information on social media during disasters, including messages with the intentional behavior of seeking or offering help. Timely identification of such help intent can operationally benefit disaster management by aiding the information collection and filtering for response planning. Prior research on intent identification has developed supervised learning methods specific to a disaster using labeled messages from that disaster. However, rapidly acquiring a large set of labeled messages is difficult during a new disaster in order to train a supervised learning classifier. In this paper, we propose a novel transfer learning method for help intent identification on Twitter during a new disaster. This method efficiently transfers the knowledge of intent behavior from the labeled messages of the past disasters using novel Sparse Coding feature representation. Our experiments using Twitter data from four disaster events show the performance gain up to 15% in both F-score and accuracy over the baseline of popular Bag-of-Words representation. The results demonstrate the applicability of our method to assist realtime help intent identification in future disasters.

Reports of the 2013 AAAI Spring Symposium Series

The Association for the Advancement of Artificial Intelligence was pleased to present the AAAI 2014 Spring Symposium Series, held Monday through Wednesday, March 24–26, 2014. The titles of the eight symposia were Applied Computational Game Theory, Big Data Becomes Personal: Knowledge into Meaning, Formal Verification and Modeling in Human-Machine Systems, Implementing Selves with Safe Motivational Systems and Self-Improvement, The Intersection of Robust Intelligence and Trust in Autonomous Systems, Knowledge Representation and Reasoning in Robotics, Qualitative Representations for Robots, and Social Hacking and Cognitive Security on the Internet and New Media). This report contains summaries of the symposia, written, in most cases, by the cochairs of the symposium.

Intent Mining for the Good, Bad, and Ugly Use of Social Web: Concepts, Methods, and Challenges

The social web has empowered us to easily share information, express opinions, and engage in discussions on events around the world. While users of social media platforms often offer help and emotional support to others (the good), they also spam (the bad) and harass others as well as even manipulate others via fake news (the ugly). In order to both leverage the positive effects and mitigate the negative effects of using social media, intent mining provides a computational approach to proactively analyze social media data. This chapter introduces an intent taxonomy of social media usage with examples and describes methods and future challenges to mine the intentional uses of social media.

Generic architecture of a social media-driven intervention support system for smart cities

With the growing adoption of Internet and mobile technology, the information exchange via social media has greatly influenced both government and corporate operations. Social Media has not only become a platform for mere entertainment and communication but a great source of innovation for public services. While millions of users generate data on these social media platforms everyday, the challenge is to effectively extract and analyze information from the big social data in a productive manner for improving public services of future smart and connected communities. In this paper, we propose a novel approach of creating an Intervention Support System (ISS) interface for public services of a city to easily and effectively monitor attitude trends of public for topics of interest (e.g., a cleanliness awareness campaign), while hiding all the complex functionality of collecting, processing, and mining big user-generated data from social media. We first discuss the generic architecture of ISS and its various components, followed by demonstrating the efficacy of the proposed architecture via an application design to identify intervention targets on social networks for supporting public campaigns against the key societal crisis of gender-based violence. We conclude with the challenges, limitations, and future work direction to effectively assist future smart city services via big data analytics approaches.