Search Procedure: Step 5
In Step 5, our tasks are to decide optimal strings based on the result from Step 4, and the databases to be used.
The search string selection and formation is a case-by-case process, but usually we use following priority order:
- F-term classification, plus wide IPC classes/full text key word strings;
- F-Index/ F-I facets classification, plus wide IPC classes/full text key word strings;
- IPC classification, plus wide F-term classes/full text key word strings;
In most cases, we do not use classification only because we consider that the possibility of miss-classification by Japanese patent examiners is quite high. We often use combination of classification and full text key word strings. For example, if we search for power management for dynamic memory, we would use following F-term classification:
(5B024 AA01+ 5B024CA23+ 5B024CA00+ 5B024DA20).
Plus we would use wide IPC classes/full text key words combination with strings such as:
(G11C11? + G11B05?) x (電源+電池+パワー)
For certain technologies, full text key word strings would be more appropriate. For example, if we need to find a cell phone with a proximity sensor, we would consider a full text key word first because the term of “proximity sensor” in Japanese is very unique; almost everybody would use the same Japanese term to describe this feature. However, because of high noise of full text key words, we usually would use either a wide F-term class or wide F-index, or wide IPC class as a filter to reduce the noise.
For patent classification, we often use PATOLIS database. For full text, we prefer NRI database considering the database functionality and precision. After the search, we combine the hits from PATOLIS and NRI, and use a tool to eliminate the repetitions.
If the full text search produces too high a noise, we may use claims and abstracts, but usually we do not use titles and abstracts, because we consider such searches are too narrow.