Findings from these activities

In Year One, members of the multimedia resource summarization team have started the collection of a large corpus of medical articles. This includes compiling a list of electronically available, relevant journals, locating full text articles, and downloading them onto a local server for subsequent post-processing for PERSIVAL. The library team provided information on journals. We provided of a list of cardiology journals published and then a list of the same journals ranked by order of importance and influence in the field of Cardiology. A presentation of these findings was given with explanation of the methodology used to accumulate the list. The summarization team has taken this information to guide the corpus creation. A high school intern has been trained to select and build the collection, as part of this effort. In the next stage of the project, we will work with this team to iteratively guide and participate in the corpus creation process to establish a reliable and trusted source which can be confidently used as a base gold standard for evaluation.

In the search area, the libraries team has provided information on specific query techniques for the OVID database. The summarization team is experimenting with structured ways to expand the patient record-vs.-article matching scenario, which is currently still quite restricted (article selection out of a small set of 200), by running patient-related, highly specific OVID queries. Hands on training for Medline and Science Citation Index was provided by the library team. Assistance in Medline (Ovid) searching and MeSH vocabulary was also provided. In year two, we will work directly with researchers to further review the way search strategies are used with highly structured metadata in order to refine queries. We will interact actively with the research teams to ensure that our search expertise is reflected in the PERSIVAL system.

In the area of consumer health, we have given an orientation to consumer health materials in our library in paper format and have recommended consumer health libraries in the area to browse the collection. We have also reviewed the Columbia Home Medical Guide, an excellent resource since there is no copyright problem. The second year of the project will involve more consumer health focus, which will enable us to contribute to this area even more actively.

In the near future, we will help the user interface team by providing literature on the "reference interview" - how it is structured, what it is trying to accomplish, how you do it, etc. in the library literature. This information on how to go from "I want something on heart disease" to "I want something on the left ventricular malfunction in older patients with diabetes and high blood pressure." We will provide information on ways that reference librarians interview to help users narrow search in order to assist the team in modeling this process within PERSIVAL. We will also be actively involved in iterative stages of the user interface construction in order to give feedback as user needs specialists. Finally, we will provide information on query styles to help construction of search mechanisms which reflect the diversity of user types.