Semi-automatically building a knowledge base of dietary nutritional information for different medical conditions
For years, health articles and dietary advice have been provided by dietitians and clinical nutritionists on different online media. This information is widely accessible through the internet. However, self-diagnosis can cause a number of problems, especially for those untrained in the field of heal...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | text |
| Language: | English |
| Published: |
Animo Repository
2015
|
| Online Access: | https://animorepository.dlsu.edu.ph/etd_bachelors/12120 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | De La Salle University |
| Language: | English |
| Summary: | For years, health articles and dietary advice have been provided by dietitians and clinical nutritionists on different online media. This information is widely accessible through the internet. However, self-diagnosis can cause a number of problems, especially for those untrained in the field of health care. Thus, this poses a need for specialized systems that can effectively relay this information to the general public. Numerous ontologies have been built to address this issue. However, the ontologies in human nutrition currently in existence are built with general use and preventive health maintenance in mind. There is currently no ontology that maps the nutritional aspects of medical conditions to different food items. In addition, the task of populating an ontology is very tedious to do by hand.
This research focuses on the design and development of a system that can semi-automatically populate a knowledge base, in the form of an ontology, which associates the necessary nutrients for medical conditions to food items that contain them. This is done by means of information extraction from various health articles available on the internet. One of the key characteristics of an ontology is its reusability. The knowledge base populated by this system is meant to be used by future systems in the field of health informatics. Based on the results of testing, the system is able to extract instances from online articles at an average precision of 0.7804, recall of 0.5149 and f-measure of 0.5149. The relationships between these instances are also mapped and represented via an ontology. An API has been provided to facilitate access to this populated ontology. |
|---|