You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Annotation prediction is constructed using the decorator pattern.
Prediction should be able to:
Search by property value
Search by property type and property value
For 1 and 2 additional option to boost certain sources
For 1 and 2 additional option to filter exclusively certain sources
For 1 and 2 additional option to boost certain topics
For 1 and 2 additional option to filter exclusively certain topics
For 3-6 should be able to rank topics or sources
If nothing found in Zooma should fall back to OLS strict search and give max GOOD confidence
If nothing returns should do a loose OLS search and return results with MEDIUM or LOW confidence
Prediction and score should be done as below (search on Solr annotation summaries):
Search
search and retrieve documents based on text search, boost most recent documents
Primary Scoring
normalise Solr score result and boost by:
source number
votes (number of documents in annotation summary)
Filter
boost sources or (if wanted exclusive sources) just filter out the unwanted ones
Secondary Scoring
Boost scores based on text similarity (Needleman-Wunsch Scorer)
Confidence can then be calculated based on cutoff scores and cutoff percentage.
The text was updated successfully, but these errors were encountered:
Annotation prediction is constructed using the decorator pattern.
Prediction should be able to:
Search by property value
Search by property type and property value
For 1 and 2 additional option to boost certain sources
For 1 and 2 additional option to filter exclusively certain sources
For 1 and 2 additional option to boost certain topics
For 1 and 2 additional option to filter exclusively certain topics
For 3-6 should be able to rank topics or sources
If nothing found in Zooma should fall back to OLS strict search and give max GOOD confidence
If nothing returns should do a loose OLS search and return results with MEDIUM or LOW confidence
Prediction and score should be done as below (search on Solr annotation summaries):
Search
search and retrieve documents based on text search, boost most recent documents
Primary Scoring
normalise Solr score result and boost by:
source number
votes (number of documents in annotation summary)
Filter
boost sources or (if wanted exclusive sources) just filter out the unwanted ones
Secondary Scoring
Boost scores based on text similarity (Needleman-Wunsch Scorer)
Confidence can then be calculated based on cutoff scores and cutoff percentage.
The text was updated successfully, but these errors were encountered: