Alana Platt

Discovering relationships among categories using misclassification information

Knowledge of relationships among categories is of the interest in different domains such as text classification, content analysis, and text mining. We propose and evaluate approaches to effectively identify relationships among document categories. Our proposed novel method capitalizes on the misclassification results of a text classifier to identify potential relationships among categories. We demonstrate that our system detects such relationships, even those relationships that assessors failed to identify in manual evaluation. Furthermore, we favorably compare the effectiveness of our methods with the state of art method and demonstrate a significant improvement in precision (34%) and recall (5%).

FACT: Fast Algorithm for Categorizing Text

With the ever-increasing number of digital documents, the ability to automatically classify those documents both quickly and accurately is becoming more critical and difficult. We present Fast Algorithm for Categorizing Text (FACT), which is a statistical based multi-way classifier with our proposed feature selection, Ambiguity Measure (AM), which uses only the most unambiguous keywords to predict the category of a document. Our empirical results show that FACT outperforms the best results on the best performing feature selection for the Naive Bayes classifier namely, Odds Ratio. We empirically show the effectiveness of our approach in outperforming Odds Ratio using four benchmark datasets with a statistical significance of 99% confidence level. Furthermore, the performance of FACT is comparable or better than current non-statistical based classifiers.

From Earthquakes to "#morecowbell": Identifying Sub-topics in Social Network Communications

Social networking websites have become a vital means of communication that can provide information on various topics. The real time nature of the information published on social networking websites coupled with their accessibility as a publishing platform make them a powerful tool for information gathering. Furthermore, many individuals utilize these sorts of platforms to share their knowledge and opinions with others. While this information is useful, there exist a number of challenges to effectively use social networks to gather information about a topic. We propose a methodology that automatically detects sub-topics and groups the social networking messages accordingly.

Organization of social network messages to improve understanding of an evolving crisis

In this work, we present a system that is implemented on top of a stream of social network messages to facilitate learning about an emerging crisis. This method automatically detects sub-topics of the topic (crisis), and populates the sub-topics with the relevant retrieved messages. In future work, we intend to reduce the repetitiveness of the generated sub-topics while maintaining high precision in our classification, and implementing other unsupervised learning versions of our method and comparing them to the KNN version.

On using user query sequence to detect off-topic search

Retrieving off-topic documents to a users pre-defined area of interest via a search engine is potentially a violation of access rights and is a concern to every private, commercial, and governmental organization. We improve content-based off-topic search detection approaches by using a sequence of user queries versus the individual queries. In this approach, we reevaluate how off-topic a query is, based on the sequence of queries that preceded it. Our empirical results show that using the information from the queries in a given query window, the false alarm rate is reduced by a statistically significant amount.

Understanding user's needs in information gathering in social networks

Social Network Sites (SNSs) have become an important method for information exchange, and many people turn to these sites for their information needs. Users encounter a variety of challenges when performing information gathering on SNSs because the SNS messages are not organized like traditional documents. These challenges include small amounts of information in each message, unorganized messages, and lack of context in each message. In this work we focus specifically on how the information should be presented to a user to address these issues. We perform a study of SNS users to identify their preferences for such an interface and present the results.

Short Query Sequences in Misuse Detection

Retrieving documents that are off-topic to an authorized users pre-defined area of interest via a search engine is potentially a violation of access rights, commonly referred to as misuse, and is a concern to every private, commercial, and governmental organization. In this preliminary study, we improve previous content-based misuse detection approaches by using a sequence of queries versus only individual user queries to perform our misuse assessment. That is, we reevaluate how off-topic a query is based on the sequence of queries that preceded it. Our empirical results show that by using the information from the queries in a given query window, the degree of false alarm and undetected misuse are both statistically significantly reduced.

Improving classification based off-topic search detection via category relationships

The illegitimate access of documents by insiders (also known as off-topic search) is an increasingly prevalent and largely ignored problem. We propose an approach that uses text classification for off-topic search detection. Our empirical results indicate that off-topic search detection effectiveness improves by considering only a subset of documents that are retrieved for a given user query. Furthermore, we also show that the effectiveness of off-topic search detection improves by using the ontological information of document categories. Our empirical results demonstrate that utilizing sibling relationship information and relationships derived from misclassification information statistically significantly improves the results over the baseline in most cases.