An approach for extracting chemical data from molecular representations

摘要

The ever-increasing quantity of chemical literature necessitates the creation of automated techniques for extracting relevant information. The digital conversion process of chemical molecule representations into their corresponding computerized illustration has been evolved using numerous applications. This work is a part of our contribution to computer-aided toxicity recognition research which predicts the possible toxic side effects of drugs on the process of drug design. The current proposed approach provides essential reviews for previous related researches in the field of automated chemical information extraction and mentions our current proposed technique that is considered as an intelligent module for converting chemical molecule structures to computerized structures. Mentioned mechanisms are used to detect bonds that are represented by letters through comparison with templates database, atoms, and lines for extracting data from graphs. A sample of 100 chemical compound structures was used to be converted into computer representation to get an overall result of 88.9% precision. Finally, a comparison between related approaches and the current proposed one at their precision rates for classification of substructure patterns was conducted.