lflage commited on
Commit
c66e748
·
verified ·
1 Parent(s): 04bb160

Create README

Browse files
Files changed (1) hide show
  1. README.md +79 -0
README.md ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Wiki-Talks
2
+
3
+ The Wiki-Talks dataset is a collection of conversational threads extracted from the talk pages on Wikipedia.
4
+ This dataset captures collaborative dialogue, discussion patterns, and consensus-building among Wikipedia contributors.
5
+ It is useful for NLP research focused on dialogue, sentiment analysis, and community dynamics.
6
+
7
+ ## Details
8
+
9
+ ### Description
10
+
11
+ The Wiki-Talks dataset contains discussion threads from Wikipedia’s talk pages.
12
+ Each thread includes a conversation title, a main message, and nested replies, capturing the sequence and structure of discussions.
13
+ This dataset provides a unique view into how contributors collaborate, negotiate meaning, and resolve conflicts when editing content on Wikipedia.
14
+
15
+ - **Curated by:** @lflage
16
+ - **Language(s) (NLP):** Various languages, reflecting the multilingual nature of Wikipedia (e.g., English, Spanish, German, and many others)
17
+ - **License:** CC BY-SA 3.0
18
+
19
+ ### Dataset Sources [optional]
20
+
21
+ - **Repository:** [Wiki-Talks](https://github.com/lflage/wiki-talks)
22
+
23
+ ## Uses
24
+
25
+ The Wiki-Talks dataset is intended for research and development in NLP fields that require structured conversational data.
26
+ Some use cases include:
27
+
28
+ - **Dialogue Modeling:** Training models for conversational agents that need to capture turn-taking and structured discourse.
29
+ - **Sentiment Analysis:** Analyzing language use, tone, and sentiment in collaborative or contentious settings.
30
+ - **Toxicity and Moderation:** Developing models for detecting inappropriate content and toxicity in user interactions.
31
+ - **Discourse and Pragmatics Research:** Studying patterns of collaboration, argumentation, and consensus-building in online communities.
32
+
33
+ ## Dataset Structure
34
+
35
+ The Wiki-Talks dataset is organized in a nested format to capture the hierarchical structure of conversations. Each thread contains:
36
+
37
+ The `threads` column contains the nested thread structure.
38
+
39
+ - `title` (str): The title of the discussion thread.
40
+ - `message` (str): The main message initiating the conversation.
41
+ - `replies` (list of reply objects): Each `reply` object contains:
42
+ - `message` (str): The reply message.
43
+ - `replies` (list of nested reply objects): Nested replies within the conversation, capturing deeper levels of response and discussion.
44
+
45
+ ```
46
+ thread = { "title": str,
47
+ "message": str,
48
+ "replies: list[reply]}
49
+ reply = { "message": str,
50
+ replies: list[reply]}
51
+ ```
52
+
53
+ ## Dataset Creation
54
+
55
+ ### Source Data
56
+
57
+ The data was collected from publicly available Wikipedia talk pages, more specifically the pages-meta-current.
58
+
59
+ #### Data Collection and Processing
60
+
61
+ Data was selected from Wikipedia's talk pages and processed to structure the threads and replies hierarchically.
62
+ The dataset includes only messages from public Wikipedia dumps and was processed using WikiTextParser, lxml and python standard libraries.
63
+ It uses the time signature of a comment on a thread to delimit a comment.
64
+
65
+ ### Recommendations
66
+
67
+ When using this dataset, the following should be considered:
68
+
69
+ - **Bias**: As Wikipedia is user-generated, discussions may reflect certain biases based on the contributors' demographics, interests, or viewpoints.
70
+ - **Limitations**: Conversations may sometimes be off-topic or contain incomplete responses due to the open nature of Wikipedia.
71
+ - **Technical**: The dataset can be large and memory-intensive due to nested structures; consider streaming methods for processing.
72
+
73
+ ## Citation [optional]
74
+
75
+ For citation please refer to the original repository.
76
+
77
+ ## Contribution
78
+
79
+ Refer to the [repository](https://github.com/lflage/wiki-talks) for how to contribute to expand the dataset