Related Experiment Videos
Comparing the information content of two large olfactory databases
Marco Pintore1, Christophe Wechman, Gilles Sicard
1BioChemics Consulting, Orléans, France.
Journal of Chemical Information and Modeling
|January 24, 2006
Summary
Subjective olfactory descriptions create discrepancies in odor databases. Adaptive fuzzy partition models help evaluate database quality and identify the most reliable odor information.
Area of Science:
- Chemistry
- Data Science
- Sensory Science
Background:
- Expert subjectivity in olfactory descriptions leads to significant discrepancies in odor profiles across different databases.
- Comparing the "Perfumery Materials and Performance 2001" (PMP2001) database with Arctander's seminal works (1960, 1969) highlights these inconsistencies.
Purpose of the Study:
- To develop a method for assessing the quality and trustworthiness of olfactory databases.
- To address the problem of discrepancies in odor profile data due to subjective expert descriptions.
Main Methods:
- Classification models were developed using the adaptive fuzzy partition method.
- Models were trained on subsets of data from the PMP2001 database and Arctander's books, categorized into identical olfactory classes.
Main Results:
- The developed models demonstrated robustness and predictive power.
- These metrics provide a quantitative basis for evaluating the information content of olfactory databases.
Conclusions:
- The adaptive fuzzy partition method offers a reliable criterion for assessing the quality of olfactory databases.
- This approach aids in determining the most trustworthy sources for odor profile information.