dfki-nlp commited on
Commit
add7897
1 Parent(s): ba61064

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -4
README.md CHANGED
@@ -67,10 +67,14 @@ task_ids:
67
  - **Size of the generated dataset:** 40.9 MB
68
  - **Total amount of disk used:** 103.2 MB
69
  ### Dataset Summary
70
- Note: This dataset reader is intended for a JSONL version of the TACRED dataset that is used internally by DFKI SLT.
71
- We use the JSONL reader because it is more efficient as a generator than fully parsing the JSON files of the
72
- original TACRED dataset. You can find the TACRED dataset reader for the JSON version of
73
- the dataset [here](https://huggingface.co/datasets/DFKI-SLT/tacred).
 
 
 
 
74
 
75
  The TAC Relation Extraction Dataset (TACRED) is a large-scale relation extraction dataset with 106,264 examples built over newswire and web text from the corpus used in the yearly TAC Knowledge
76
  Base Population (TAC KBP) challenges. Examples in TACRED cover 41 relation types as used in the TAC KBP challenges (e.g., per:schools_attended
@@ -81,6 +85,8 @@ Note: There is currently a label-corrected version of the TACRED dataset, which
81
  the original version released in 2017. For more details on this new version, see the [TACRED Revisited paper](https://aclanthology.org/2020.acl-main.142/)
82
  published at ACL 2020.
83
 
 
 
84
  ### Supported Tasks and Leaderboards
85
  - **Tasks:** Relation Classification
86
  - **Leaderboards:** [https://paperswithcode.com/sota/relation-extraction-on-tacred](https://paperswithcode.com/sota/relation-extraction-on-tacred)
 
67
  - **Size of the generated dataset:** 40.9 MB
68
  - **Total amount of disk used:** 103.2 MB
69
  ### Dataset Summary
70
+ NOTE: This Datasetreader supports a reduced version of the original TACRED JSON format with the following changes:
71
+ - Removed fields: stanford_pos, stanford_ner, stanford_head, stanford_deprel, docid
72
+ The motivation for this is that we want to support additional languages, for which these fields were not required
73
+ or available. The reader expects the specification of a language-specific configuration specifying the variant
74
+ (original or revised) and the language (as a two-letter iso code). The default config is 'original-en'.
75
+
76
+ You can find the TACRED dataset reader for the original version of the
77
+ dataset [here](https://huggingface.co/datasets/DFKI-SLT/tacred).
78
 
79
  The TAC Relation Extraction Dataset (TACRED) is a large-scale relation extraction dataset with 106,264 examples built over newswire and web text from the corpus used in the yearly TAC Knowledge
80
  Base Population (TAC KBP) challenges. Examples in TACRED cover 41 relation types as used in the TAC KBP challenges (e.g., per:schools_attended
 
85
  the original version released in 2017. For more details on this new version, see the [TACRED Revisited paper](https://aclanthology.org/2020.acl-main.142/)
86
  published at ACL 2020.
87
 
88
+
89
+
90
  ### Supported Tasks and Leaderboards
91
  - **Tasks:** Relation Classification
92
  - **Leaderboards:** [https://paperswithcode.com/sota/relation-extraction-on-tacred](https://paperswithcode.com/sota/relation-extraction-on-tacred)