• 1
• 2
• 3
• 4
• 5
• 6
• 7
33
Text-mining methods used 
for information extraction 
in plant scientific papers
4. Entity categorization
34CC-BY
Categorization
Different entities predicted could be similar 
to others, and be linked to a single 
category.
Categorization allows to extract the 
information and to go back to a superior 
level of conceptualization compared to the 
form of the raw text.
35CC-BY
Categorization
In the case of synonyms: explains that AP2 
and APETALA 2 are the same concept, and 
could be linked to a unique category such as 
AT4G36920 (locus of the gene) 
➢ Improves the query of TDM results
➢ Allows link with external resources
36CC-BY
Ontology for Entity recognition & 
Categorization
To detect and categorize some complex 
entities such as Development Phase or 
Tissue, we use ontologies. 
An ontology is a semantic representation of 
knowledge. Concepts are organized and 
linked together by different relations, for 
example "is_a", "part_of", "has_part" ...
37CC-BY
Example of an ontology
Plant Ontology
We can reason 
and infer knowledge 
from ontologies.
38CC-BY
A plant ontology for TDM use
It will be built to allow detections of 
complex entities, such as Tissue, 
Developmental Phase, Pathway ... 
It will be used for :
● Entity detection
● Categorization of predicted entities 
● The interrogation of the final application