Abstract:
This paper investigates the potential use of dialog-based ALICEbots in disseminating terrorism information to the general public. In particular, we study the acceptance and response satisfaction of ALICEbot responses in both the general conversation and terrorism domains. From our analysis of three different knowledge sets: general conversation, terrorism, and combined, we found that users were more favorable to the systems that exhibited conversational flow. We also found that the system that incorporated both conversation and terrorism knowledge performed better than systems with only conversation or terrorism knowledge alone. Lastly, we were interested in what types of questions were the most prevalently used and discovered that questions beginning with 'wh*' words were the most popular method to start an interrogative sentence. However, 'wh* sentence starters surprisingly proved to be in a very narrow majority. © 2006 Elsevier B.V. All rights reserved.
Abstract:
Temporal information extraction from clinical narratives is of critical importance to many clinical applications. We participated in the EVENT/TIMEX3 track of the 2012 i2b2 clinical temporal relations challenge, and presented our temporal information extraction system, MedTime. MedTime comprises a cascade of rule-based and machine-learning pattern recognition procedures. It achieved a micro-averaged f-measure of 0.88 in both the recognitions of clinical events and temporal expressions. We proposed and evaluated three time normalization strategies to normalize relative time expressions in clinical texts. The accuracy was 0.68 in normalizing temporal expressions of dates, times, durations, and frequencies. This study demonstrates and evaluates the integration of rule-based and machine-learning-based approaches for high performance temporal information extraction from clinical narratives. © 2013 Elsevier Inc.
Abstract:
Quantitative assessments of researchers' productivity and impacts are crucial for the information systems (IS) discipline. Motivated by its growing popularity and expanding use, we offer a perspective on the h index, which refers to the number of papers a researcher has coauthored with at least h citations each. We studied a partial list of 232 top IS researchers who received doctoral degrees between 1957 and 2003 and chose Google Scholar as the source for our analyses. At the individual level, we attempted to identify some of the most productive, high-impact researchers, as well as those who exhibited impressive paces of productivity. At the institution level, we revealed some institutions with relatively more productive researchers, as well as institutions that had produced more productive researchers. We also analyzed the overall IS community by examining the primary research areas of productive scholars identified by our analyses. We then compared their h index scores with those of top scholars in several related disciplines. © 2011 ACM.
Abstract:
As the key technique to build domain-specific search engines, focused crawling has drawn a lot of attention from researchers in the past decade. However, as Web structure analysis techniques advance, several problems in traditional focused crawler design were revealed and they could result in domain-specific collections with low quality. In this work, we studied the problems of focused crawling that are caused by using local search algorithms. We also proposed to use a global search algorithm, the Genetic Algorithm, in focused crawling to address the problems. We conducted evaluation experiments to examine the effectiveness of our approach. The results showed that our approach could build domain-specific collections with higher quality than traditional focused crawling techniques. Furthermore, we used the concept of Web communities to evaluate how comprehensively the focused crawlers could traverse the Web search space, which could be a good complement to the traditional focused crawler evaluation methods.
