readme
Browse files
README.md
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
---
|
2 |
language:
|
3 |
- en
|
4 |
-
|
5 |
- NER
|
6 |
- named entity recognition
|
7 |
- RE
|
@@ -9,13 +9,20 @@ language:
|
|
9 |
- entity mention detection
|
10 |
- EMD
|
11 |
- coreference resolution
|
12 |
-
|
13 |
-
|
14 |
- Ontonotes
|
15 |
- CoNLL04
|
16 |
---
|
17 |
|
18 |
# CoReNer
|
19 |
|
20 |
-
|
21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
language:
|
3 |
- en
|
4 |
+
tags:
|
5 |
- NER
|
6 |
- named entity recognition
|
7 |
- RE
|
|
|
9 |
- entity mention detection
|
10 |
- EMD
|
11 |
- coreference resolution
|
12 |
+
license: apache-2.0
|
13 |
+
datasets:
|
14 |
- Ontonotes
|
15 |
- CoNLL04
|
16 |
---
|
17 |
|
18 |
# CoReNer
|
19 |
|
20 |
+
## Model description
|
21 |
+
|
22 |
+
A multi-task model for named-entity recognition, relation extraction, entity mention detection and coreference resolution.
|
23 |
+
|
24 |
+
We model NER as a span classification task, and relation extraction as a multi-label classification of (NER) span tuples.
|
25 |
+
Similarly, model EMD as a span classification task and CR as a binary classification of (EMD) span tuples.
|
26 |
+
To construct the CR clusters, we keep the top antecedent of each mention, then compute the connected components of the mentions' undirected graph.
|
27 |
+
|
28 |
+
See additional details and usage examples at: https://github.com/aiola-lab/corener.
|