High Accuracy Retrieval from Documents (HARD)
| Metadata | Value |
|---|---|
| Purpose | Background: the searcher wants to know where the topic came from |
| Details: the searcher wants to know the details of the topic | |
| Answer: the user is looking for an answer to a specific question | |
| Any: the user has no specific purpose in mind or has not specified one | |
| Genre | Overview: the searcher is interested in general news related to the topic |
| Reaction: the searcher is looking for news commentary on the topic | |
| I-Reaction: is like REACTION but is specifically about non-U.S. news commentary | |
| Administrative: the searcher is interested in official US government documents | |
| Any: any genre is acceptable or none was indicated | |
| Familiarity | 1 : no prior knowledge |
| 2 - 4: increasing familiarity | |
| 5: know details of topic | |
| Unknown: the user does not know his or her familiarity or has not specified one | |
| Granularity | Document: the searcher is expecting complete documents (one or more) |
| Passage: the searcher expects extracts from documents that are on the paragraph or multi-paragraph level | |
| Sentence: the retrieved units should be roughly at the sentence level | |
| Phrase: the user is expecting a small number of words (including just one) as a response | |
| Any: means the user has no specific granularity in mind or did not specify one | |
| Related Text | "...": identified text (not from the corpus) that is known to be related to the topic being specified. This provides a kind of pre-query relevance feedback |
Back to Main HARD Project page