Data mining of fractured experimental data using neurofuzzy logic-discovering and integrating knowledge hidden in...

Area of Science:

• Pharmaceutical Science
• Chemical Engineering
• Data Mining

Background:

• Current pharmaceutical process understanding relies on single experimental databases, limiting knowledge extraction.
• Fractured data from multiple sources contain interrelated information crucial for complex processes like fluid-bed granulation.

Purpose of the Study:

• To investigate data mining strategies for discovering and integrating knowledge from multiple small experimental databases.
• To apply neurofuzzy logic technology to enhance knowledge discovery in pharmaceutical processes.
• To improve the understanding of fluid-bed granulation by integrating fragmented data.

Main Methods:

• Employed three distinct data mining strategies.
• Utilized neurofuzzy logic technology for knowledge discovery.
• Integrated data from multiple small experimental databases, including textual information.

Main Results:

• More comprehensive domain knowledge was discovered by integrating multiple databases using appropriate data mining strategies.
• Textual information, often excluded, proved critical for integrating knowledge across databases.
• Generic knowledge of the fluid-bed granulation domain was successfully extracted.

Conclusions:

• Integrating fractured data via advanced data mining strategies significantly enhances pharmaceutical process understanding.
• Neurofuzzy logic is effective for discovering and integrating knowledge from disparate data sources.
• Consideration of all data types, including textual information, is essential for comprehensive domain knowledge extraction.