Fitts' law specifies a logarithmic relationship between motion duration and the ratio of target distance over target size. This paper introduces two large open-access datasets from experimental user studies: first a controlled (in-lab) study with 46 participants, and second an uncontrolled online study using a Java applet. We present a succinct derivation of the square-root variant of Fitts' law using optimal control theory and compare three models that relate motion duration to the ratio of target distance over target size: LOG (Fitts' original logarithmic function), SQR (square-root), and LOG' (McKenzie's logarithmic plus 1.0). We find that: 1) the data from the controlled and uncontrolled studies are consistent; 2) for homogeneous targets (with fixed size and distance), the SQR model yields a significantly better fit than LOG or LOG', except with the most difficult targets (where the ratio of target distance over target size is large) where the models are not significantly different; and 3) for heterogeneous targets (with varying size and distance), SQR yields a significantly better fit than LOG for easy targets and LOG yields a significantly better fit for targets of medium difficulty, while the LOG' model yields a significantly better fit than both LOG and SQR on very difficult targets. The anonymized datasets including 94 580 human reaching motion timing measurements are, to our knowledge, the largest collected to date.

Standard Sentiment Analysis applies Natural Language Processing methods to assess an "approval" value of a given text, categorizing it into "negative", "neutral", or "positive" or on a linear scale. Sentiment Analysis can be used to infer ratings values for users based on textual reviews of items such as books, films, or products. We propose an approach to generalizing the concept to multiple dimensions to estimate user ratings along multiple axes such as "service", "price" and "value". We use Canonical Correlation Analysis (CCA) and derive a mathematical model that can be used as a multivariate regression tool. This model has a number of valuable properties: it can be trained offline and used efficiently on live stream of texts like blogs and tweets, can be used for visualization and data clustering and labeling, and finally it can potentially be incorporated into natural language product search algorithms. At the end we propose an evaluation procedure that can be used on live data when a ground truth is not available. Based on this model we present our preliminary results from empirical data that we have collected from our system Opinion Space1. We show that for this dataset the CCA model outperforms the PCA that was originally used in Opinion Space.

Internet users are increasingly inclined to contribute comments to online news articles, videos, product reviews, and blogs. The most common interface for comments is a list, sorted by time of entry or by binary ratings. It is widely recognized that such lists do not scale well and can lead to "cyberpolarization," which serves to reinforce extreme opinions. We present Opinion Space: a new online interface incorporating ideas from deliberative polling, dimensionality reduction, and collaborative filtering that allows participants to visualize and navigate through a diversity of comments. This self-organizing system automatically highlights the comments found most insightful by users from a range of perspectives. We report results of a controlled user study. When Opinion Space was compared with a chronological List interface, participants read a similar diversity of comments. However, they were significantly more engaged with the system, and they had significantly higher agreement with and respect for the comments they read.