MoodViews: Tools for Blog Mood Analysis
AAAI Symposium on Computational Approaches to Analysing Weblogs (AAAI-CAAW), page 153--154. (2006)Abstract
About MoodViews
MoodViews is developed by the Moodteam, a group of researchers working on information access—search, retrieval, and discovery, especially of web information. From the point of view of information access, the blogspace offers many natural opportunities beyond traditional search facilities, such as trend detection, topic tracking, link tracking, feed generation, etc. But there is more. Many blog authoring environments allow bloggers to tag their entries with highly individual (and personal) features. Users of LiveJournal, currently the largest weblog community, have the option of reporting their "mood" at the time of the post; users can either select a mood from a predefined list of 132 common moods such as ämused'' or ängry,'' or enter free-text.
A large percentage of LiveJournal users chooses to utilize this option, tagging their postings with a mood. This results in a stream of hundreds of weblog posts tagged with mood information per minute, from hundreds of thousands of different users across the globe. Our focus in this demo is on providing access to the blogspace using moods as the "central'' dimension. The type of information needs that we are interested in are best illustrated by questions such as: How do moods develop? How are they related? How do global events impact moods? And: Can global mood swings be traced back to global events?
MoodViews is a collection of tools for analyzing, tracking and visualizing moods and mood changes in blogs posted by LiveJournal users.
MoodViews tracks the stream of mood-annotated text made available by LiveJournal. At present, MoodViews consists of three components, each offering a different view of global mood levels, the aggregate across all postings of the various moods: Moodgrapher tracks the global mood levels, Moodteller predicts them, and Moodsignals helps in understanding the underlying reasons for mood changes.
Moodgrapher
MoodViews builds on Moodgrapher, the basic component of the system, which plots the aggregate mood levels over time. Sample plots, showing irregular mood patterns following events with global implications, are shown on this page. Moodgrapher was launched in June 2005. Concept and design: Gilad Mishne. Implementation: Gilad Mishne and Krisztian Balog.
Moodteller
Moodteller goes a step beyond Moodgrapher, and uses natural language processing and machine learning to estimate the mood levels from the text of blog entries posted on LiveJournal, without using the mood tags provided by bloggers. The estimation is then plotted and compared to the actual values based on tags provided by bloggers, and accuracy information is reported. Moodteller was launched in September 2005. Concept, design and implementation: Gilad Mishne.
Moodsignals
Moodsignals detects words and phrases which are associated with a given mood in a given time interval, using statistical frequency comparisons and burstiness models. Moodsignals was launched in March 2006. Concept: Gilad Mishne. Design: Gilad Mishne and Krisztian Balog. Implementation: Krisztian Balog and Breyten Ernsting.
Moodspotter
Given a topic, Moodspotter tells you which moods are typically (or: were recently) associated with that topic. A beta version of Moodspotter was launched in early 2007. Concept: Krisztian Balog and Maarten de Rijke. Design: Krisztian Balog and Maarten de Rijke. Implementation: Krisztian Balog and Breyten Ernsting.
Moodsearch
In response to a query, Moodsearch ranks blog posts by time, relevance, or "mood". A beta version is planned for the second half of 2007. Concept: Krisztian Balog and Maarten de Rijke. Design: Kirsztian Balog and Maarten de Rijke. Implementation: Svetlin Stefanov, Krisztian Balog and Breyten Ernsting.
Links and resources
Tags
community
@huiyangsfsu's tags highlighted