Connecting the Ephemeral and Archival Information Networks
Activities and Results
Major Activities:
UMass has made significant progress on new lines of research that has grown out of our earlier work on the interaction between ephemeral and archival information. We have continued to work on models for retrieval and summarization that incorporate features from social media. This work is a collaboration with the RMIT group mentioned in the original proposal. We have also been vigorously pursuing the approach of integrating social and archival media through answer passage retrieval. In this approach, social media is used as both training data and a source of answers for longer, “non-factoid” questions. This approach to integration through question answering has led us to neural models, which resulted in some delays due to the learning curve. We have also had to develop new testbeds for this research, including scaling up the number of question and answer pairs available for training through using social media collections and crowdsourcing. The neural model research, although a new direction, is based on the original hypothesis for the proposal - searching either ephemeral or archival information will be enhanced using the connections between them.
Key Outcomes or Other Achievements:
The key outcomes for UMass were the new publications, and the models that they describe. The outcomes for CMU were the new publications, results on improving social search, and improved crawling algorithms.
This work is supported in part by the Center for Intelligent Information Retrieval (CIIR) and in part by the National Science Foundation (NSF IIS-1160894 and NSF IIS-1160862).
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.