Commit
•
5f38d5c
1
Parent(s):
3508d6a
Update README.md
Browse files
README.md
CHANGED
|
@@ -39,7 +39,7 @@ Similarity Calculation: Uses precomputed embeddings and cosine similarity to ran
|
|
| 39 |
Efficiency: Designed to handle large volumes of BPMN files and queries efficiently.
|
| 40 |
|
| 41 |
## Shortcomings:
|
| 42 |
-
Data Availability: Lack of sufficient BPMN files within ABB, we trained on the data of [camunda/bpmn-for-research](https://github.com/camunda/bpmn-for-research) by converting the BPMN files to Networkx graphs and measuring the Minimum Edit Distance between random pairs.
|
| 43 |
Minimum Edit Distance: **Does this translate to perceived similarity?** It might require a different approach. Training data in the form of (BPMN, BPMN, similarity) is needed. If user interactions can be captured, this might be a better approach for gauging perceived similarity or general 'useful' similarity.
|
| 44 |
|
| 45 |
## Future Steps:
|
|
|
|
| 39 |
Efficiency: Designed to handle large volumes of BPMN files and queries efficiently.
|
| 40 |
|
| 41 |
## Shortcomings:
|
| 42 |
+
Data Availability: Lack of sufficient BPMN files within ABB, we trained on the data of [camunda/bpmn-for-research](https://github.com/camunda/bpmn-for-research) by converting the BPMN files to Networkx graphs and measuring the Minimum Edit Distance between random pairs.
|
| 43 |
Minimum Edit Distance: **Does this translate to perceived similarity?** It might require a different approach. Training data in the form of (BPMN, BPMN, similarity) is needed. If user interactions can be captured, this might be a better approach for gauging perceived similarity or general 'useful' similarity.
|
| 44 |
|
| 45 |
## Future Steps:
|