Grant: $71,200 - National Institutes of Health - Sep. 1, 2009
No votes have been cast for this award yet
Award Description: The long-term objective of the proposed project is to provide the public health (PH) workforce with improved access to high quality, relevant PH grey literature reports in order to positively impact the planning, conducting, and evaluating of PH interventions. The project will consist of two components: 1) continuation of user-focused technical system development, and 2) deployment and evaluation of this system's impact on the tasks of the PH workforce in county health departments. The system will automatically harvest web-based grey literature reports (utilizing rules trained on input from PH professionals) and then produce rich summaries of PH grey literature reports using the model of essential elements of PH intervention reports validated by PH specialists (Turner et al, In Press). After developing a user interface that capitalizes on both natural language querying and the display of search results in a structured, model-based summary, the system will be introduced into several county health departments and evaluated to determine its impact on information flows and uses. The project will include intrinsic evaluations of: 1) the appropriateness of the grey literature reports harvested by the system; 2) the quality and sufficiency of summaries representing the full reports based on the PH intervention model and; 3) retrieval results for users' queries using metrics of precision and recall. Extrinsic evaluation will be done in PH departments participating in the New York Academy of Medicine's ongoing study of the relationship between information and effectiveness. The research will provide baseline measures. We will evaluate: 1) the ease and frequency of use, perceptions of currency, accuracy, and completeness of retrieved reports by county PH personnel, and; 2) impact of the system's usage on information flows and uses based on internal outcome measures within the county health departments. The work proposed here shares both the missions of public health and the National Library of Medicine through the design of an information system that exploits natural language processing technology to efficiently collect and provide access to quality public health grey literature for the public health workforce.
Project Description: The purpose of the original grant was to provide improved access to public health grey literature in order to meet the information needs of the public health workforce. The purpose of the ARRA-funded Supplement is to extend and improve performance of the tools built under the parent grant. The research to be conducted will have three main contributions. First, we will substantially improve the search tool by improving underlying retrieval and ranking techniques. In particular, we expect to incorporate technical advances in query formulation and ranking. This research is expected to materially improve the quality of the rankings. Second, we will integrate NLM?s Unified Medical Language System (UMLS) into the NLP processing used for query formulation to improve synonym and near synonym expansion. Third, we will evaluate and implement techniques for expanding the effective size of the collection by improving filtering and by evaluating the effectiveness of federating our targeted results with results from other search engines. Lastly, we will carry out a user evaluation similar to the Workplace Evaluation done prior to the improvements, allowing us to compare results and better document the impact of the changes made. Milestones and objectives: 1. Improve Underlying Retrieval and Ranking: a) Develop test collection (fix document collection, collect queries and relevance judgments), b) Compare baseline ranking with Lemur ranking, c) Evaluate fusion strategies, and d) Generate pseudo feedback expansion alternatives and test; 2. Integrate UMLS into Query Formulation: a) Generate query expansion strategies, and b) Compare expanded query performance with baseline; 3. Expand Effective Size of Collection: a) Improve crawl filter, expand crawl scope, evaluate effect on results, and b) Evaluate effectiveness of federating multiple results. Evaluation: The project period began Sept 30, the last day of the reporting period.
Jobs Summary: Not yet applicable. (Total jobs reported: 0)
Project Status: Not Started
This award's data was last updated on Sep. 1, 2009. Help expand these official descriptions using the wiki below.