Rohan Kumar commited on
Commit
b6a0133
·
verified ·
1 Parent(s): ec81012

feat: Files are ready to upload

Browse files
README.md CHANGED
@@ -1,3 +1,278 @@
1
  ---
2
  license: cc-by-nc-4.0
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc-by-nc-4.0
3
+ configs:
4
+ - config_name: main_updated_dataset
5
+ data_files: "scored_sample_cut_new.json"
6
+ default: true
7
+ - config_name: main_dataset
8
+ data_files: "scored_sample_cut.json"
9
  ---
10
+
11
+
12
+ # Beyond N-Grams (BNG)
13
+
14
+
15
+
16
+ ## Dataset Summary
17
+ This dataset is designed to study **language trends over time**. It combines multi-source metadata, aggregated review signals, and LLM-generated features to enable hypothesis-driven and exploratory research on how narrative and reception characteristics evolve over time.
18
+
19
+ The dataset does not contain original book text. Instead, it uses **LLM-generated proxy content and derived features** to approximate semantic and evaluative properties.
20
+
21
+ ---
22
+
23
+ ## Dataset Description
24
+
25
+ ### Construction Process
26
+ The dataset is built using a three-stage pipeline:
27
+
28
+ 1. **Data Collection**
29
+ - Book metadata and summaries collected from:
30
+ - Google Books
31
+ - Wikipedia
32
+ - Goodreads
33
+ - Project Gutenberg
34
+ - Includes publication details, author attributes, ratings, and review statistics
35
+
36
+ 2. **Proxy Content Generation**
37
+ - LLM-generated content used to approximate book semantics
38
+ - Based on research-backed prompting methods
39
+
40
+ 3. **Feature Engineering**
41
+ - ~100 structured features per book, including:
42
+ - Metadata features
43
+ - Review-based signals
44
+ - LLM-derived evaluative scores
45
+
46
+ ---
47
+
48
+ ## Dataset Structure
49
+
50
+ - Each row represents a **single book**
51
+ - Includes:
52
+ - Book metadata
53
+ - Author metadata
54
+ - Publication year
55
+ - Aggregated review statistics
56
+ - Derived feature set
57
+ - Some fields may contain nested structures (e.g., reviews)
58
+
59
+ ---
60
+
61
+ ## Data Fields
62
+
63
+ The dataset consists of structured features grouped into three main categories: metadata, review-based signals, and LLM-generated evaluative attributes.
64
+
65
+ ---
66
+
67
+ ### Metadata Features
68
+
69
+ - `book_title`: Book title/name
70
+ - `book_author`: Book author name
71
+ - `gutenberg_book_link`: Publically accessible book link
72
+ - `preview_text`: Publically available book text
73
+ - `reconstructed_text`: Proxy text of the book generated by the model
74
+ - `author_degree_level`: Highest known academic qualification of the author
75
+ - `author_gender`: Gender of the author
76
+ - `author_nationality`: Nationality of the author
77
+ - `pub_year`: Year of publication
78
+ - `genre_count`: Number of genres associated with the book
79
+ - `genres`: List of genres
80
+
81
+ ---
82
+
83
+ ### Review and Engagement Features
84
+
85
+ - `review_count`: Number of reviews included
86
+ - `reviews`: Nested list containing:
87
+ - `reviewer_review_count`: Total reviews by reviewer (if available)
88
+ - `reviewer_follower_count`: Follower count of reviewer
89
+ - `rating_out_of_5`: Rating given in the review
90
+ - `review_year`: Year of the review
91
+ - `likes`: Number of likes on the review
92
+ - `review_length`: Length of the review
93
+
94
+ - `total_reviews`: Total number of reviews
95
+ - `total_ratings`: Total number of ratings
96
+ - `average_rating`: Average rating score
97
+ - `early_avg_rating`: Early-stage average rating (if available)
98
+ - `star_1_count` to `star_5_count`: Distribution of ratings across 1–5 stars
99
+
100
+ ---
101
+
102
+ ### LLM-Generated Features
103
+
104
+
105
+ The dataset includes ~100 LLM-generated features grouped into the following categories:
106
+
107
+ ---
108
+
109
+ #### 1. Sentiment and Polarity
110
+ Includes features measuring overall sentiment, intensity, shifts, and implicit emotional signals.
111
+ Examples: `polarity_score`, `sentiment_intensity`, `aspect_based_sentiment`, `sentiment_shift`, `implicit_sentiment`, `mixed_sentiment_ratio`
112
+
113
+ ---
114
+
115
+ #### 2. Emotion and Affect
116
+ Captures emotional states based on Ekman categories and affective dynamics.
117
+ Examples: `ekman_anger_intensity`, `ekman_joy_intensity`, `ekman_sadness_intensity`, `ekman_fear_intensity`, `ekman_surprise_intensity`, `affective_arousal`, `emotional_variance`
118
+
119
+ ---
120
+
121
+ #### 3. Sarcasm, Irony, and Figurative Language
122
+ Measures figurative expression, ambiguity, and creative language use.
123
+ Examples: `sarcasm_probability`, `situational_irony`, `hyperbole_intensity`, `linguistic_creativity`, `idiomatic_density`, `interpretation_flexibility`, `ambiguity_score`
124
+
125
+ ---
126
+
127
+ #### 4. Political Ideology and Populism
128
+ Represents ideological stance and group dynamics.
129
+ Examples: `left_right_leaning`, `authoritarian_libertarian`, `populist_rhetoric_intensity`, `anti_elitism_score`, `in_group_favoritism`, `out_group_derogation`
130
+
131
+ ---
132
+
133
+ #### 5. Social Media and Cultural Signals
134
+ Captures informal communication and cultural context markers.
135
+ Examples: `emoji_sentiment_alignment`, `hashtag_relevance`, `slang_density`, `code_mixing_ratio`, `meme_reference_likelihood`, `cultural_resonance`, `acronym_frequency`
136
+
137
+ ---
138
+
139
+ #### 6. Readability, Cohesion, and Cognitive Load
140
+ Measures complexity, structure, and interpretability of content.
141
+ Examples: `cognitive_load_score`, `information_density`, `semantic_entropy`, `surprisal_score`, `novelty_score`, `conceptual_depth`, `abstraction_level`, `context_dependence`, `self_containment`, `attention_retention_curve`
142
+
143
+ ---
144
+
145
+ #### 7. Argumentation and Epistemology
146
+ Captures logical structure, claims, and evidence.
147
+ Examples: `argument_polarity`, `causal_density`, `claim_verifiability`, `evidence_implicitness`, `claim_explicitness`, `evidence_strength`, `logical_coherence`
148
+
149
+ ---
150
+
151
+ #### 8. Subjectivity and Stance
152
+ Represents opinion strength and perspective diversity.
153
+ Examples: `opinionatedness`, `subjectivity_score`, `stance_towards_topic`, `direct_opinion_presence`, `implicit_stance_cues`, `perspective_multiplicity`
154
+
155
+ ---
156
+
157
+ #### 9. Style, Formality, and Tone
158
+ Captures writing style and tonal consistency.
159
+ Examples: `temporal_drift`, `tone_stability`, `hook_strength`, `memory_salience`, `formality_score`, `academic_tone_intensity`, `conversational_fluidity`, `authoritative_tone`
160
+
161
+ ---
162
+
163
+ #### 10. Toxicity, Controversy, and Discourse
164
+ Measures conversational impact and conflict potential.
165
+ Examples: `discursive_engagement`, `quote_worthiness`, `reader_alignment`, `controversy_potential`, `toxicity_score`, `insult_intensity`
166
+
167
+ ---
168
+
169
+ #### 11. Factuality, Misinformation, and Credibility
170
+ Evaluates truthfulness and information reliability.
171
+ Examples: `truth_alignment_score`, `sensationalism_index`, `rumor_stance`, `source_attribution_clarity`, `clickbait_potential`, `fabrication_likelihood`
172
+
173
+ ---
174
+
175
+ #### 12. Persuasion and Propaganda
176
+ Captures rhetorical influence and manipulation strategies.
177
+ Examples: `pathos_appeal`, `ethos_appeal`, `loaded_language_density`, `fear_mongering_index`, `glittering_generality_score`, `logical_fallacy_presence`, `call_to_action_strength`
178
+
179
+ ---
180
+
181
+ #### 13. Dialogue Acts and Conversational Intent
182
+ Represents communicative intent and interaction signals.
183
+ Examples: `information_seeking_intent`, `agreement_score`, `disagreement_intensity`, `directive_intent_score`, `commissive_intent`, `acknowledgment_presence`
184
+
185
+ ---
186
+
187
+ #### 14. Empathy and Psychological Signals
188
+ Captures emotional understanding and mental state indicators.
189
+ Examples: `reflection_trigger_score`, `cognitive_empathy_score`, `emotional_support_intensity`, `distress_indicator`, `validation_score`, `self_disclosure_depth`
190
+
191
+ ---
192
+
193
+ #### 15. Humor and Wordplay
194
+ Represents humor styles and creative expression.
195
+ Examples: `humor_intensity`, `pun_presence_score`, `absurdity_index`, `self_deprecation_level`, `dark_humor_probability`
196
+
197
+ ---
198
+
199
+
200
+ ## Data Sources
201
+
202
+ This dataset is derived from publicly accessible sources:
203
+ - Google Books, Gutenberg (metadata and summaries)
204
+ - Wikipedia (author and contextual information)
205
+ - Goodreads (aggregated review statistics)
206
+ - LLMs (proxy content and synthetic annotation)
207
+
208
+ No raw proprietary text is redistributed. Only **aggregated and derived representations** are included.
209
+
210
+ ---
211
+
212
+ ## Annotation Process
213
+
214
+ LLM-based annotation was used to generate proxy content and evaluative features:
215
+ - Prompt-based generation strategy
216
+ - Model-generated scores for 100+ features.
217
+
218
+ ### Human Validation
219
+ - 48 participants evaluated a subset of samples
220
+ - Fleiss' kappa: **0.59**
221
+
222
+ ---
223
+
224
+ ## Intended Use
225
+
226
+ This dataset is suitable for:
227
+ - Research on persuasion trends across time
228
+ - Hypothesis testing in computational social science
229
+ - Analysis of relationships between content and reception
230
+ - Studying behavior of LLM-generated features
231
+
232
+ ---
233
+
234
+ ## Out-of-Scope Use
235
+
236
+ This dataset should **not** be used for:
237
+ - Training production models
238
+ - Ground-truth benchmarking
239
+ - Author profiling or individual evaluation
240
+ - High-stakes decision-making systems
241
+ - Literary or stylistic analysis of original text
242
+
243
+ ---
244
+
245
+ ## Bias and Limitations
246
+
247
+ - Popularity bias due to reliance on review platforms
248
+ - Demographic attributes may be inferred
249
+ - LLM-generated features may reflect model bias
250
+
251
+ ---
252
+
253
+ ## Personal and Sensitive Information
254
+
255
+ - No personally identifiable information (PII)
256
+ - Data is aggregated and non-traceable to individuals
257
+ - Contains limited demographic attributes (e.g., gender, nationality)
258
+
259
+ ---
260
+
261
+ ## License
262
+
263
+ This dataset is released under **CC-BY-NC-4.0**.
264
+
265
+ Rationale:
266
+ - Derived from mixed sources including platform-restricted data
267
+ - Includes synthetic and aggregated features
268
+ - Restricted to non-commercial research use
269
+
270
+ ---
271
+
272
+ ## Ethical Considerations
273
+
274
+ - Dataset is intended for research use only
275
+ - Not suitable for decision-making involving individuals
276
+ - Potential biases and misuse risks are documented
277
+ - No raw user-generated content is directly included
278
+
gitattributes ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.avro filter=lfs diff=lfs merge=lfs -text
4
+ *.bin filter=lfs diff=lfs merge=lfs -text
5
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
6
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
7
+ *.ftz filter=lfs diff=lfs merge=lfs -text
8
+ *.gz filter=lfs diff=lfs merge=lfs -text
9
+ *.h5 filter=lfs diff=lfs merge=lfs -text
10
+ *.joblib filter=lfs diff=lfs merge=lfs -text
11
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
12
+ *.lz4 filter=lfs diff=lfs merge=lfs -text
13
+ *.mds filter=lfs diff=lfs merge=lfs -text
14
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
15
+ *.model filter=lfs diff=lfs merge=lfs -text
16
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
17
+ *.npy filter=lfs diff=lfs merge=lfs -text
18
+ *.npz filter=lfs diff=lfs merge=lfs -text
19
+ *.onnx filter=lfs diff=lfs merge=lfs -text
20
+ *.ot filter=lfs diff=lfs merge=lfs -text
21
+ *.parquet filter=lfs diff=lfs merge=lfs -text
22
+ *.pb filter=lfs diff=lfs merge=lfs -text
23
+ *.pickle filter=lfs diff=lfs merge=lfs -text
24
+ *.pkl filter=lfs diff=lfs merge=lfs -text
25
+ *.pt filter=lfs diff=lfs merge=lfs -text
26
+ *.pth filter=lfs diff=lfs merge=lfs -text
27
+ *.rar filter=lfs diff=lfs merge=lfs -text
28
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
29
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
30
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
31
+ *.tar filter=lfs diff=lfs merge=lfs -text
32
+ *.tflite filter=lfs diff=lfs merge=lfs -text
33
+ *.tgz filter=lfs diff=lfs merge=lfs -text
34
+ *.wasm filter=lfs diff=lfs merge=lfs -text
35
+ *.xz filter=lfs diff=lfs merge=lfs -text
36
+ *.zip filter=lfs diff=lfs merge=lfs -text
37
+ *.zst filter=lfs diff=lfs merge=lfs -text
38
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
39
+ # Audio files - uncompressed
40
+ *.pcm filter=lfs diff=lfs merge=lfs -text
41
+ *.sam filter=lfs diff=lfs merge=lfs -text
42
+ *.raw filter=lfs diff=lfs merge=lfs -text
43
+ # Audio files - compressed
44
+ *.aac filter=lfs diff=lfs merge=lfs -text
45
+ *.flac filter=lfs diff=lfs merge=lfs -text
46
+ *.mp3 filter=lfs diff=lfs merge=lfs -text
47
+ *.ogg filter=lfs diff=lfs merge=lfs -text
48
+ *.wav filter=lfs diff=lfs merge=lfs -text
49
+ # Image files - uncompressed
50
+ *.bmp filter=lfs diff=lfs merge=lfs -text
51
+ *.gif filter=lfs diff=lfs merge=lfs -text
52
+ *.png filter=lfs diff=lfs merge=lfs -text
53
+ *.tiff filter=lfs diff=lfs merge=lfs -text
54
+ # Image files - compressed
55
+ *.jpg filter=lfs diff=lfs merge=lfs -text
56
+ *.jpeg filter=lfs diff=lfs merge=lfs -text
57
+ *.webp filter=lfs diff=lfs merge=lfs -text
58
+ # Video files - compressed
59
+ *.mp4 filter=lfs diff=lfs merge=lfs -text
60
+ *.webm filter=lfs diff=lfs merge=lfs -text
gpt-5-feature-annotation-prompt-template.txt ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ prompt = f"""You are a social scientist conducting a rigorous content analysis study.
2
+
3
+ Score the following text on the dimension of "{dimension}".
4
+ Definition of this dimension: {desc}
5
+
6
+ General scoring guide (interpret the endpoints according to the definition above):
7
+ - 0–2: The dimension is essentially absent, or strongly present at its lowest/negative extreme.
8
+ - 3–4: The dimension is weakly present, or present with clear mitigating/contrary elements.
9
+ - 5: Neutral, ambiguous, mixed, or the text gives no clear evidence either way.
10
+ - 6–7: The dimension is moderately to clearly present.
11
+ - 8–10: The dimension is strongly and unambiguously present at its highest/positive extreme.
12
+
13
+ Instructions:
14
+ - Base your score strictly on evidence in the text itself. Do not use outside knowledge, assumptions about the topic, or stereotypes about the subject matter to infer a score.
15
+ - If the text contains no relevant evidence for this dimension, score it 5.
16
+ - If evidence is mixed, weigh by prominence and intensity in the text, not merely by counting instances.
17
+ - Do not let text length, formatting, or verbosity bias the score — judge content, not volume.
18
+ - Judge only what is explicitly present or strongly implied; do not over-interpret.
19
+ - Return ONLY a JSON array containing a single integer from 0 to 10. No explanation, no reasoning, no markdown formatting, no additional keys or text.
20
+
21
+ Example output: [6]
22
+
23
+ Text: {chunk}
24
+ """
scored_sample_cut.json ADDED
The diff for this file is too large to render. See raw diff