Keyword elicitation for patent retrieval by using bibliographic information

Keyword selection is one of the most important tasks for patent retrieval. However, few researchers have focused on how to choose keywords appropriately in comparison with to improve retrieval performance via techniques from Bibliometics, such as patent counts, citation and so on. The paper has proposed, thus, a new keyword collection algorithm. The algorithm adopts a snowball approach that elicits rarely used keywords from commonly used keywords to build a technical term set under the assistance of patent experts. The elicitation process was inspired by the phenomenon that some inventors tend to use special keywords called design-around synonyms in their issued patents in order to elude the search. Applying the algorithm to DVD3C and DVD6C patent pools, the experiments indicated that those design around synonyms would be found successfully.