Update README.md
Browse files
README.md
CHANGED
@@ -136,9 +136,6 @@ The documents are retained in their original file formats without parsing or seg
|
|
136 |
|
137 |
### Dataset Description
|
138 |
|
139 |
-
<!-- Provide a longer summary of what this dataset is. -->
|
140 |
-
|
141 |
-
|
142 |
|
143 |
- **Curated by:** Yulong Hui, Tsinghua University
|
144 |
- **Language(s) (NLP):** English
|
@@ -166,7 +163,20 @@ Assess the effectiveness of parsing strategies on unstructured PDF documents.
|
|
166 |
## Dataset Structure
|
167 |
|
168 |
<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
169 |
|
|
|
170 |
Field Name | Field Value | Description| Example
|
171 |
--- | --- | ---|---
|
172 |
doc_name | string | name of the source document | 1912.01214
|
@@ -174,7 +184,7 @@ q_uid | string | unique id of the question | 9a05a5f4351db75da371f7ac12eb0b
|
|
174 |
question | string | raised question | which datasets did they experiment with?
|
175 |
answer <br />or answer_1, answer_2 <br />or short_answer, long_answer | string | ground truth answer/answers | Europarl, MultiUN
|
176 |
|
177 |
-
**Additional Notes:**
|
178 |
|
179 |
|
180 |
## Dataset Creation
|
|
|
136 |
|
137 |
### Dataset Description
|
138 |
|
|
|
|
|
|
|
139 |
|
140 |
- **Curated by:** Yulong Hui, Tsinghua University
|
141 |
- **Language(s) (NLP):** English
|
|
|
163 |
## Dataset Structure
|
164 |
|
165 |
<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
|
166 |
+
#### Descriptive Statistics
|
167 |
+
|
168 |
+
Sub Dataset (folder_name) | Source Domain | Doc Format | Doc Num | Q&A Num | Avg #Words | Avg #Pages |Q&A Types
|
169 |
+
--- | --- | --- | --- | --- | ---| --- |---
|
170 |
+
FinHybrid (fin) | finance reports | PDF | 788 | 8190 | 76.6k | 147.8 |arithmetic
|
171 |
+
TatHybrid (tat)| finance reports |PDF|170|14703|77.5k|148.5|extractive, counting, arithmetic
|
172 |
+
PaperTab (paper_tab)| academic papers |PDF|307|393|6.1k|11.0|extractive, yes/no, free-form
|
173 |
+
PaperText (paper_text)| academic papers | PDF|1087|2804|5.9k|10.6|extractive, yes/no, free-form
|
174 |
+
FetaTab (feta)| wikipedia |PDF & HTML|878|1023|6.0k|14.9|free-form
|
175 |
+
NqText (nq)| wikipedia |PDF & HTML|645|2477|6.1k|14.9|extractive
|
176 |
+
|
177 |
+
|
178 |
|
179 |
+
#### Data Fields
|
180 |
Field Name | Field Value | Description| Example
|
181 |
--- | --- | ---|---
|
182 |
doc_name | string | name of the source document | 1912.01214
|
|
|
184 |
question | string | raised question | which datasets did they experiment with?
|
185 |
answer <br />or answer_1, answer_2 <br />or short_answer, long_answer | string | ground truth answer/answers | Europarl, MultiUN
|
186 |
|
187 |
+
**Additional Notes:** Some sub-datasets may have multiple ground_truth answers, where the answers are organized as `answer_1`, `answer_2` (in FinHybrid, PaperTab and PaperText) or `short_answer`, `long_answer` (in NqText); In sub-dataset TatHybrid, the answer is organized as a sequence, due to the involvement of the multi-span Q&A type. Additionally, some sub-datasets may have unique data fields. For example, `doc_url` in FetaTab and NqText describes the Wikipedia URL page, while `answer_type` and `answer_scale` in TatHybrid provide extended answer references.
|
188 |
|
189 |
|
190 |
## Dataset Creation
|