anonymous220217 commited on
Commit
ae910dc
·
verified ·
1 Parent(s): d2aa0e5

Upload 6 files

Browse files
README.md CHANGED
@@ -1,3 +1,274 @@
1
  ---
2
  license: cc-by-nc-4.0
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc-by-nc-4.0
3
+ configs:
4
+ - config_name: booksemA
5
+ data_files: "booksemA.json"
6
+ default: true
7
+ - config_name: booksemB
8
+ data_files: "booksemB.json"
9
  ---
10
+
11
+
12
+ # BookSem
13
+
14
+
15
+ ## Dataset Summary
16
+ 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.
17
+
18
+ The dataset does not contain original book text. Instead, it uses **LLM-generated proxy content and derived features** to approximate semantic and evaluative properties.
19
+
20
+ ---
21
+
22
+ ## Dataset Description
23
+
24
+ ### Construction Process
25
+ The dataset is built using a three-stage pipeline:
26
+
27
+ 1. **Data Collection**
28
+ - Book metadata and summaries collected from:
29
+ - Google Books
30
+ - Wikipedia
31
+ - Goodreads
32
+ - Project Gutenberg
33
+ - Includes publication details, author attributes, ratings, and review statistics
34
+
35
+ 2. **Proxy Content Generation**
36
+ - LLM-generated content used to approximate book semantics
37
+ - Based on research-backed prompting methods
38
+
39
+ 3. **Feature Engineering**
40
+ - ~118 structured features per book, including:
41
+ - Metadata features
42
+ - Review-based signals
43
+ - LLM-derived evaluative scores
44
+
45
+ ---
46
+
47
+ ## Dataset Structure
48
+
49
+ - Each row represents a **single book**
50
+ - Includes:
51
+ - Book metadata
52
+ - Author metadata
53
+ - Publication year
54
+ - Aggregated review statistics
55
+ - Derived feature set
56
+ - Some fields may contain nested structures (e.g., reviews)
57
+
58
+ ---
59
+
60
+ ## Data Fields
61
+
62
+ The dataset consists of structured features grouped into three main categories: metadata, review-based signals, and LLM-generated evaluative attributes.
63
+
64
+ ---
65
+
66
+ ### Metadata Features
67
+
68
+ - `book_title`: Book title/name
69
+ - `book_author`: Book author name
70
+ - `gutenberg_book_link`: Publically accessible book link
71
+ - `preview_text`: Publically available book text
72
+ - `reconstructed_text`: Proxy text of the book generated by the model
73
+ - `author_degree_level`: Highest known academic qualification of the author
74
+ - `author_gender`: Gender of the author
75
+ - `author_nationality`: Nationality of the author
76
+ - `pub_year`: Year of publication
77
+ - `genre_count`: Number of genres associated with the book
78
+ - `genres`: List of genres
79
+
80
+ ---
81
+
82
+ ### Review and Engagement Features
83
+
84
+ - `review_count`: Number of reviews included
85
+ - `reviews`: Nested list containing:
86
+ - `reviewer_review_count`: Total reviews by reviewer (if available)
87
+ - `reviewer_follower_count`: Follower count of reviewer
88
+ - `rating_out_of_5`: Rating given in the review
89
+ - `review_year`: Year of the review
90
+ - `likes`: Number of likes on the review
91
+ - `review_length`: Length of the review
92
+
93
+ - `total_reviews`: Total number of reviews
94
+ - `total_ratings`: Total number of ratings
95
+ - `average_rating`: Average rating score
96
+ - `early_avg_rating`: Early-stage average rating (if available)
97
+ - `star_1_count` to `star_5_count`: Distribution of ratings across 1–5 stars
98
+
99
+ ---
100
+
101
+ ### LLM-Generated Features
102
+
103
+
104
+ The dataset includes ~118 LLM-generated features grouped into the following categories:
105
+
106
+ ---
107
+
108
+ #### 1. Sentiment and Polarity
109
+ Includes features measuring overall sentiment, intensity, shifts, and implicit emotional signals.
110
+ Examples: `polarity_score`, `sentiment_intensity`, `aspect_based_sentiment`, `sentiment_shift`, `implicit_sentiment`, `mixed_sentiment_ratio`, `emotional_valence`, `optimistic_outlook`
111
+
112
+ ---
113
+
114
+ #### 2. Emotion and Affect
115
+ Captures emotional states based on Ekman categories and affective dynamics.
116
+ Examples: `ekman_anger_intensity`, `ekman_joy_intensity`, `ekman_sadness_intensity`, `ekman_fear_intensity`, `ekman_surprise_intensity`, `affective_arousal`, `emotional_variance`, `emotional_intensity`
117
+
118
+ ---
119
+
120
+ #### 3. Sarcasm, Irony, and Figurative Language
121
+ Measures figurative expression, ambiguity, and creative language use.
122
+ Examples: `sarcasm_probability`, `situational_irony`, `hyperbole_intensity`, `linguistic_creativity`, `idiomatic_density`, `interpretation_flexibility`, `ambiguity_score`
123
+
124
+ ---
125
+
126
+ #### 4. Political Ideology and Populism
127
+ Represents ideological stance and group dynamics.
128
+ Examples: `left_right_leaning`, `authoritarian_libertarian`, `populist_rhetoric_intensity`, `anti_elitism_score`, `in_group_favoritism`, `out_group_derogation`, `individualism_collectivism`, `social_transformation`, `hierarchical_worldview`, `tradition_preservation`
129
+
130
+ ---
131
+
132
+ #### 5. Social Media and Cultural Signals
133
+ Captures informal communication and cultural context markers.
134
+ Examples: `emoji_sentiment_alignment`, `hashtag_relevance`, `slang_density`, `code_mixing_ratio`, `meme_reference_likelihood`, `cultural_resonance`, `acronym_frequency`, `gender_orientation`, `feminine_home_work_association`, `masculine_home_work_association`
135
+
136
+ ---
137
+
138
+ #### 6. Readability, Cohesion, and Cognitive Load
139
+ Measures complexity, structure, and interpretability of content.
140
+ Examples: `cognitive_load_score`, `information_density`, `semantic_entropy`, `surprisal_score`, `novelty_score`, `conceptual_depth`, `abstraction_level`, `context_dependence`, `self_containment`, `attention_retention_curve`, `conceptual_complexity`
141
+
142
+ ---
143
+
144
+ #### 7. Argumentation and Epistemology
145
+ Captures logical structure, claims, and evidence.
146
+ Examples: `argument_polarity`, `causal_density`, `claim_verifiability`, `evidence_implicitness`, `claim_explicitness`, `evidence_strength`, `logical_coherence`, `scientific_thinking`, `rational_thinking`, `moral_certainty`
147
+
148
+ ---
149
+
150
+ #### 8. Subjectivity and Stance
151
+ Represents opinion strength and perspective diversity.
152
+ Examples: `opinionatedness`, `subjectivity_score`, `stance_towards_topic`, `direct_opinion_presence`, `implicit_stance_cues`, `perspective_multiplicity`
153
+
154
+ ---
155
+
156
+ #### 9. Style, Formality, and Tone
157
+ Captures writing style and tonal consistency.
158
+ Examples: `temporal_drift`, `tone_stability`, `hook_strength`, `memory_salience`, `formality_score`, `academic_tone_intensity`, `conversational_fluidity`, `authoritative_tone`
159
+
160
+ ---
161
+
162
+ #### 10. Toxicity, Controversy, and Discourse
163
+ Measures conversational impact and conflict potential.
164
+ Examples: `discursive_engagement`, `quote_worthiness`, `reader_alignment`, `controversy_potential`, `toxicity_score`, `insult_intensity`
165
+
166
+ ---
167
+
168
+ #### 11. Factuality, Misinformation, and Credibility
169
+ Evaluates truthfulness and information reliability.
170
+ Examples: `truth_alignment_score`, `sensationalism_index`, `rumor_stance`, `source_attribution_clarity`, `clickbait_potential`, `fabrication_likelihood`, `evidence_based_reasoning`
171
+
172
+ ---
173
+
174
+ #### 12. Persuasion and Propaganda
175
+ Captures rhetorical influence and manipulation strategies.
176
+ Examples: `pathos_appeal`, `ethos_appeal`, `loaded_language_density`, `fear_mongering_index`, `glittering_generality_score`, `logical_fallacy_presence`, `call_to_action_strength`, `threat_and_fear_appeal`, `persuasive_intent`
177
+
178
+ ---
179
+
180
+ #### 13. Dialogue Acts and Conversational Intent
181
+ Represents communicative intent and interaction signals.
182
+ Examples: `information_seeking_intent`, `agreement_score`, `disagreement_intensity`, `directive_intent_score`, `commissive_intent`, `acknowledgment_presence`
183
+
184
+ ---
185
+
186
+ #### 14. Empathy and Psychological Signals
187
+ Captures emotional understanding and mental state indicators.
188
+ Examples: `reflection_trigger_score`, `cognitive_empathy_score`, `emotional_support_intensity`, `distress_indicator`, `validation_score`, `self_disclosure_depth`, `empathetic_perspective`
189
+
190
+ ---
191
+
192
+ #### 15. Humor and Wordplay
193
+ Represents humor styles and creative expression.
194
+ Examples: `humor_intensity`, `pun_presence_score`, `absurdity_index`, `self_deprecation_level`, `dark_humor_probability`
195
+
196
+ ---
197
+
198
+
199
+ ## Data Sources
200
+
201
+ This dataset is derived from publicly accessible sources:
202
+ - Google Books, Gutenberg (metadata and summaries)
203
+ - Wikipedia (author and contextual information)
204
+ - Goodreads (aggregated review statistics)
205
+ - LLMs (proxy content and synthetic annotation)
206
+
207
+ No raw proprietary text is redistributed. Only **aggregated and derived representations** are included.
208
+
209
+ ---
210
+
211
+ ## Annotation Process
212
+
213
+ LLM-based annotation was used to generate proxy content and evaluative features:
214
+ - Prompt-based generation strategy
215
+ - Model-generated scores for 118+ features.
216
+
217
+ ### Human Validation
218
+ - 48 participants evaluated a subset of samples
219
+ - Fleiss' kappa: **0.59**
220
+
221
+ ---
222
+
223
+ ## Intended Use
224
+
225
+ This dataset is suitable for:
226
+ - Research on persuasion trends across time
227
+ - Hypothesis testing in computational social science
228
+ - Analysis of relationships between content and reception
229
+ - Studying behavior of LLM-generated features
230
+
231
+ ---
232
+
233
+ ## Out-of-Scope Use
234
+
235
+ This dataset should **not** be used for:
236
+ - Training production models
237
+ - Ground-truth benchmarking
238
+ - Author profiling or individual evaluation
239
+ - High-stakes decision-making systems
240
+ - Literary or stylistic analysis of original text
241
+
242
+ ---
243
+
244
+ ## Bias and Limitations
245
+
246
+ - Popularity bias due to reliance on review platforms
247
+ - Demographic attributes may be inferred
248
+ - LLM-generated features may reflect model bias
249
+
250
+ ---
251
+
252
+ ## Personal and Sensitive Information
253
+
254
+ - No personally identifiable information (PII)
255
+ - Data is aggregated and non-traceable to individuals
256
+ - Contains limited demographic attributes (e.g., gender, nationality)
257
+
258
+ ---
259
+
260
+ ## License
261
+
262
+ This dataset is released under **CC-BY-NC-4.0**.
263
+
264
+ Rationale:
265
+ - Derived from mixed sources, including platform-restricted data
266
+ - Includes synthetic and aggregated features
267
+ - Restricted to non-commercial research use
268
+
269
+ ---
270
+
271
+ ## Ethical Considerations
272
+
273
+ - No raw user-generated content is directly included
274
+
booksemA.json ADDED
The diff for this file is too large to render. See raw diff
 
booksemB.json ADDED
The diff for this file is too large to render. See raw diff
 
data.json ADDED
The diff for this file is too large to render. See raw diff
 
features_118.json ADDED
@@ -0,0 +1,592 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "category": "Sentiment and Polarity",
4
+ "dimension": "polarity_score",
5
+ "desc": "The overall positive-to-negative valence of the text. A score of 0 indicates overwhelmingly negative sentiment, 5 indicates neutral/balanced sentiment, and 10 indicates overwhelmingly positive sentiment expressed through evaluative words, judgments, and framing."
6
+ },
7
+ {
8
+ "category": "Sentiment and Polarity",
9
+ "dimension": "sentiment_intensity",
10
+ "desc": "The strength or magnitude of whichever sentiment is expressed, regardless of direction. Low scores reflect mild, understated, or flat emotional tone; high scores reflect forceful, emphatic, or extreme expression of feeling (positive or negative)."
11
+ },
12
+ {
13
+ "category": "Sentiment and Polarity",
14
+ "dimension": "aspect_based_sentiment",
15
+ "desc": "The degree to which sentiment is clearly and specifically attached to particular entities, features, or aspects mentioned in the text (e.g., a product's battery life, a policy's cost) rather than being diffuse or attached to the text as a whole."
16
+ },
17
+ {
18
+ "category": "Sentiment and Polarity",
19
+ "dimension": "sentiment_shift",
20
+ "desc": "The extent to which sentiment changes, pivots, or transitions in direction or intensity over the course of the text. Low scores indicate static, unchanging sentiment; high scores indicate clear turns (e.g., from positive to negative)."
21
+ },
22
+ {
23
+ "category": "Sentiment and Polarity",
24
+ "dimension": "implicit_sentiment",
25
+ "desc": "The degree to which sentiment is conveyed indirectly through tone, framing, connotation, or context rather than explicit evaluative words (e.g., 'great' or 'terrible'). High scores mean sentiment must be inferred; low scores mean sentiment is stated plainly or absent."
26
+ },
27
+ {
28
+ "category": "Sentiment and Polarity",
29
+ "dimension": "mixed_sentiment_ratio",
30
+ "desc": "The extent to which the text contains a genuine, simultaneous mixture of positive and negative sentiment (ambivalence) rather than a single dominant polarity. High scores indicate strong co-presence of both positive and negative evaluations."
31
+ },
32
+ {
33
+ "category": "Sentiment and Polarity",
34
+ "dimension": "sentiment_consistency",
35
+ "desc": "The degree to which the sentiment expressed remains stable, coherent, and non-contradictory throughout the text. High scores indicate a uniform emotional stance; low scores indicate erratic or self-contradictory sentiment."
36
+ },
37
+ {
38
+ "category": "Emotion and Affect",
39
+ "dimension": "ekman_anger_intensity",
40
+ "desc": "The degree to which the text expresses or evokes anger, as indicated by hostile language, frustration, indignation, or aggressive framing, following Ekman's basic emotion of anger."
41
+ },
42
+ {
43
+ "category": "Emotion and Affect",
44
+ "dimension": "ekman_joy_intensity",
45
+ "desc": "The degree to which the text expresses or evokes joy, happiness, delight, or enthusiasm, following Ekman's basic emotion of joy."
46
+ },
47
+ {
48
+ "category": "Emotion and Affect",
49
+ "dimension": "ekman_sadness_intensity",
50
+ "desc": "The degree to which the text expresses or evokes sadness, grief, disappointment, or sorrow, following Ekman's basic emotion of sadness."
51
+ },
52
+ {
53
+ "category": "Emotion and Affect",
54
+ "dimension": "ekman_fear_intensity",
55
+ "desc": "The degree to which the text expresses or evokes fear, anxiety, worry, or threat perception, following Ekman's basic emotion of fear."
56
+ },
57
+ {
58
+ "category": "Emotion and Affect",
59
+ "dimension": "ekman_surprise_intensity",
60
+ "desc": "The degree to which the text expresses or evokes surprise, astonishment, or unexpectedness, following Ekman's basic emotion of surprise."
61
+ },
62
+ {
63
+ "category": "Emotion and Affect",
64
+ "dimension": "affective_arousal",
65
+ "desc": "The general level of emotional activation or energy conveyed by the text, independent of valence. Low scores indicate calm, low-energy, subdued tone; high scores indicate excited, agitated, or highly activated emotional expression."
66
+ },
67
+ {
68
+ "category": "Emotion and Affect",
69
+ "dimension": "emotional_variance",
70
+ "desc": "The degree to which multiple distinct emotions (rather than a single dominant one) are expressed or fluctuate across the text, reflecting emotional complexity or volatility."
71
+ },
72
+ {
73
+ "category": "Sarcasm, Irony, and Figurative Language",
74
+ "dimension": "sarcasm_probability",
75
+ "desc": "The likelihood that the text uses sarcasm — saying the opposite of what is meant, typically to mock, criticize, or express contempt, often signaled by exaggeration or incongruity between literal meaning and context."
76
+ },
77
+ {
78
+ "category": "Sarcasm, Irony, and Figurative Language",
79
+ "dimension": "situational_irony",
80
+ "desc": "The degree to which the text describes or expresses irony arising from an outcome that contrasts sharply with what would normally be expected, independent of sarcastic tone."
81
+ },
82
+ {
83
+ "category": "Sarcasm, Irony, and Figurative Language",
84
+ "dimension": "hyperbole_intensity",
85
+ "desc": "The degree to which the text uses deliberate exaggeration or overstatement for emphasis or rhetorical effect, rather than literal, measured description."
86
+ },
87
+ {
88
+ "category": "Sarcasm, Irony, and Figurative Language",
89
+ "dimension": "linguistic_creativity",
90
+ "desc": "The extent to which the text uses inventive, novel, or unconventional language — wordplay, unusual metaphors, neologisms, or stylistic experimentation — rather than routine, formulaic phrasing."
91
+ },
92
+ {
93
+ "category": "Sarcasm, Irony, and Figurative Language",
94
+ "dimension": "idiomatic_density",
95
+ "desc": "The frequency and prominence of idioms, fixed expressions, and figures of speech (e.g., 'spill the beans') relative to literal, plain language."
96
+ },
97
+ {
98
+ "category": "Sarcasm, Irony, and Figurative Language",
99
+ "dimension": "interpretation_flexibility",
100
+ "desc": "The degree to which the text supports multiple plausible readings or interpretations rather than a single, fixed meaning."
101
+ },
102
+ {
103
+ "category": "Sarcasm, Irony, and Figurative Language",
104
+ "dimension": "ambiguity_score",
105
+ "desc": "The overall degree of vagueness, double meaning, or lack of clarity in the text's language, structure, or intent, such that a reader cannot easily pin down a single meaning."
106
+ },
107
+ {
108
+ "category": "Political Ideology and Populism",
109
+ "dimension": "left_right_leaning",
110
+ "desc": "The degree to which the text's rhetoric, framing, or policy positions align with a left-leaning (progressive) versus right-leaning (conservative) ideological orientation, where higher scores indicate a stronger, more explicit right-leaning stance and lower scores a stronger left-leaning stance, with 5 indicating centrist or non-ideological content."
111
+ },
112
+ {
113
+ "category": "Political Ideology and Populism",
114
+ "dimension": "authoritarian_libertarian",
115
+ "desc": "The degree to which the text favors strong central authority, order, and control (authoritarian) versus individual liberty, autonomy, and limited authority (libertarian), with higher scores reflecting stronger authoritarian framing."
116
+ },
117
+ {
118
+ "category": "Political Ideology and Populism",
119
+ "dimension": "populist_rhetoric_intensity",
120
+ "desc": "The degree to which the text frames society as a struggle between 'the pure people' and a 'corrupt elite,' using populist rhetorical strategies such as appeals to the common person and distrust of established institutions."
121
+ },
122
+ {
123
+ "category": "Political Ideology and Populism",
124
+ "dimension": "anti_elitism_score",
125
+ "desc": "The degree to which the text expresses distrust, criticism, or hostility toward elites, experts, or established institutions (political, economic, media, or academic)."
126
+ },
127
+ {
128
+ "category": "Political Ideology and Populism",
129
+ "dimension": "in_group_favoritism",
130
+ "desc": "The degree to which the text expresses preferential treatment, praise, or solidarity toward a group the author identifies with, relative to outsiders."
131
+ },
132
+ {
133
+ "category": "Political Ideology and Populism",
134
+ "dimension": "out_group_derogation",
135
+ "desc": "The degree to which the text disparages, blames, or expresses hostility toward a group perceived as outside the author's own identity group."
136
+ },
137
+ {
138
+ "category": "Social Media and Cultural Signals",
139
+ "dimension": "emoji_sentiment_alignment",
140
+ "desc": "The degree to which emojis used in the text reinforce and align with the sentiment expressed in the accompanying words, rather than contradicting or being absent from the text."
141
+ },
142
+ {
143
+ "category": "Social Media and Cultural Signals",
144
+ "dimension": "hashtag_relevance",
145
+ "desc": "The degree to which any hashtags present are topically relevant and meaningfully connected to the content of the text, rather than generic, unrelated, or absent."
146
+ },
147
+ {
148
+ "category": "Social Media and Cultural Signals",
149
+ "dimension": "slang_density",
150
+ "desc": "The frequency and prominence of informal slang, internet-speak, or colloquial vocabulary relative to standard or formal vocabulary."
151
+ },
152
+ {
153
+ "category": "Social Media and Cultural Signals",
154
+ "dimension": "code_mixing_ratio",
155
+ "desc": "The degree to which the text mixes multiple languages or dialects within the same passage, rather than remaining in a single consistent language."
156
+ },
157
+ {
158
+ "category": "Social Media and Cultural Signals",
159
+ "dimension": "meme_reference_likelihood",
160
+ "desc": "The likelihood that the text references, quotes, or alludes to internet memes, viral trends, or shared online cultural in-jokes."
161
+ },
162
+ {
163
+ "category": "Social Media and Cultural Signals",
164
+ "dimension": "cultural_resonance",
165
+ "desc": "The degree to which the text draws on or evokes shared cultural knowledge, references, traditions, or values that would resonate with a specific community or audience."
166
+ },
167
+ {
168
+ "category": "Social Media and Cultural Signals",
169
+ "dimension": "acronym_frequency",
170
+ "desc": "The frequency and prominence of acronyms and abbreviations (e.g., 'LOL', 'IMO', 'ASAP') relative to fully spelled-out language."
171
+ },
172
+ {
173
+ "category": "Readability, Cohesion, and Cognitive Load",
174
+ "dimension": "cognitive_load_score",
175
+ "desc": "The mental effort required to process and understand the text, based on sentence complexity, vocabulary difficulty, and structural density. Low scores indicate easy, effortless reading; high scores indicate demanding, effortful reading."
176
+ },
177
+ {
178
+ "category": "Readability, Cohesion, and Cognitive Load",
179
+ "dimension": "information_density",
180
+ "desc": "The amount of distinct information, facts, or ideas packed into the text relative to its length. Low scores indicate sparse, padded, or repetitive content; high scores indicate tightly packed, information-rich content."
181
+ },
182
+ {
183
+ "category": "Readability, Cohesion, and Cognitive Load",
184
+ "dimension": "semantic_entropy",
185
+ "desc": "The degree of unpredictability or diversity in the meaning and topics covered by the text. Low scores indicate narrow, repetitive, predictable semantic content; high scores indicate wide-ranging, unpredictable, or topically diverse content."
186
+ },
187
+ {
188
+ "category": "Readability, Cohesion, and Cognitive Load",
189
+ "dimension": "surprisal_score",
190
+ "desc": "The degree to which the text's word choices, claims, or turns of phrase are unexpected or improbable given the preceding context, rather than predictable and conventional."
191
+ },
192
+ {
193
+ "category": "Readability, Cohesion, and Cognitive Load",
194
+ "dimension": "conceptual_depth",
195
+ "desc": "The degree to which the text engages with complex, layered, or nuanced ideas versus simple, surface-level statements."
196
+ },
197
+ {
198
+ "category": "Readability, Cohesion, and Cognitive Load",
199
+ "dimension": "abstraction_level",
200
+ "desc": "The degree to which the text discusses abstract concepts, theories, or generalizations versus concrete, specific, tangible details."
201
+ },
202
+ {
203
+ "category": "Readability, Cohesion, and Cognitive Load",
204
+ "dimension": "context_dependence",
205
+ "desc": "The degree to which understanding the text requires outside context, prior knowledge, or surrounding discourse, rather than being fully interpretable on its own."
206
+ },
207
+ {
208
+ "category": "Readability, Cohesion, and Cognitive Load",
209
+ "dimension": "self_containment",
210
+ "desc": "The degree to which the text is a complete, self-sufficient unit of meaning that can be understood without any external context or additional information."
211
+ },
212
+ {
213
+ "category": "Readability, Cohesion, and Cognitive Load",
214
+ "dimension": "attention_retention_curve",
215
+ "desc": "The degree to which the text is structured to sustain reader attention and engagement from beginning to end, rather than losing the reader through monotony, digression, or excessive length."
216
+ },
217
+ {
218
+ "category": "Argumentation and Epistemology",
219
+ "dimension": "argument_polarity",
220
+ "desc": "The overall directional stance of the argument(s) made in the text — whether the arguments support (positive) or oppose (negative) the position or claim under discussion, with 5 indicating a balanced or neutral argumentative stance."
221
+ },
222
+ {
223
+ "category": "Argumentation and Epistemology",
224
+ "dimension": "causal_density",
225
+ "desc": "The frequency and prominence of explicit or implicit causal claims (X causes/leads to/results in Y) within the text, relative to purely descriptive or correlational statements."
226
+ },
227
+ {
228
+ "category": "Argumentation and Epistemology",
229
+ "dimension": "claim_verifiability",
230
+ "desc": "The degree to which claims made in the text are stated in a way that could, in principle, be checked or verified against external evidence, rather than being purely subjective or unfalsifiable."
231
+ },
232
+ {
233
+ "category": "Argumentation and Epistemology",
234
+ "dimension": "evidence_implicitness",
235
+ "desc": "The degree to which supporting evidence for claims is only hinted at, assumed, or left implicit rather than being explicitly stated or cited."
236
+ },
237
+ {
238
+ "category": "Argumentation and Epistemology",
239
+ "dimension": "claim_explicitness",
240
+ "desc": "The degree to which the text's central claims or assertions are stated directly and unambiguously, rather than being buried, hedged, or left for the reader to infer."
241
+ },
242
+ {
243
+ "category": "Argumentation and Epistemology",
244
+ "dimension": "evidence_strength",
245
+ "desc": "The robustness and persuasiveness of the evidence (data, examples, citations, expert testimony) provided to support the claims made in the text."
246
+ },
247
+ {
248
+ "category": "Argumentation and Epistemology",
249
+ "dimension": "logical_coherence",
250
+ "desc": "The degree to which the text's reasoning follows a clear, internally consistent logical structure without contradictions, non-sequiturs, or gaps."
251
+ },
252
+ {
253
+ "category": "Subjectivity and Stance",
254
+ "dimension": "opinionatedness",
255
+ "desc": "The overall degree to which the text expresses personal opinions, judgments, or evaluations rather than presenting neutral, purely factual content."
256
+ },
257
+ {
258
+ "category": "Subjectivity and Stance",
259
+ "dimension": "subjectivity_score",
260
+ "desc": "The degree to which the text reflects the author's personal feelings, beliefs, or interpretations rather than objective, verifiable facts."
261
+ },
262
+ {
263
+ "category": "Subjectivity and Stance",
264
+ "dimension": "stance_towards_topic",
265
+ "desc": "The clarity and strength with which the text takes a discernible position (for, against, or neutral) toward the specific topic it addresses."
266
+ },
267
+ {
268
+ "category": "Subjectivity and Stance",
269
+ "dimension": "direct_opinion_presence",
270
+ "desc": "The degree to which opinions are stated explicitly and directly (e.g., 'I believe', 'this is wrong') rather than being absent or only implied."
271
+ },
272
+ {
273
+ "category": "Subjectivity and Stance",
274
+ "dimension": "implicit_stance_cues",
275
+ "desc": "The degree to which the author's stance must be inferred from subtle cues such as word choice, framing, or emphasis rather than being explicitly declared."
276
+ },
277
+ {
278
+ "category": "Subjectivity and Stance",
279
+ "dimension": "perspective_multiplicity",
280
+ "desc": "The degree to which the text presents or acknowledges multiple viewpoints or perspectives on the topic, rather than presenting a single, one-sided view."
281
+ },
282
+ {
283
+ "category": "Style, Formality, and Tone",
284
+ "dimension": "temporal_drift",
285
+ "desc": "The degree to which the text's tone, focus, or framing shifts or evolves noticeably as it progresses, rather than remaining stylistically constant throughout."
286
+ },
287
+ {
288
+ "category": "Style, Formality, and Tone",
289
+ "dimension": "tone_stability",
290
+ "desc": "The degree to which the text maintains a single, consistent tone (e.g., serious, playful, urgent) from beginning to end without abrupt tonal shifts."
291
+ },
292
+ {
293
+ "category": "Style, Formality, and Tone",
294
+ "dimension": "hook_strength",
295
+ "desc": "The degree to which the opening of the text is crafted to immediately capture reader interest and attention, rather than beginning in a flat or unremarkable way."
296
+ },
297
+ {
298
+ "category": "Style, Formality, and Tone",
299
+ "dimension": "memory_salience",
300
+ "desc": "The degree to which the text contains distinctive, memorable phrasing, imagery, or ideas likely to stick with a reader after they finish reading."
301
+ },
302
+ {
303
+ "category": "Style, Formality, and Tone",
304
+ "dimension": "formality_score",
305
+ "desc": "The degree to which the text uses formal, polished, standard language and structure versus casual, colloquial, or conversational language."
306
+ },
307
+ {
308
+ "category": "Style, Formality, and Tone",
309
+ "dimension": "academic_tone_intensity",
310
+ "desc": "The degree to which the text adopts a scholarly, technical, or academic register, including precise terminology, hedged claims, and citation-like structure."
311
+ },
312
+ {
313
+ "category": "Style, Formality, and Tone",
314
+ "dimension": "conversational_fluidity",
315
+ "desc": "The degree to which the text reads like natural, easy spoken conversation, with informal transitions and a personal, chatty rhythm, rather than stiff or formal prose."
316
+ },
317
+ {
318
+ "category": "Style, Formality, and Tone",
319
+ "dimension": "authoritative_tone",
320
+ "desc": "The degree to which the text is written with confidence, command, and assertiveness that projects expertise and authority, rather than tentativeness or uncertainty."
321
+ },
322
+ {
323
+ "category": "Toxicity, Controversy, and Discourse",
324
+ "dimension": "discursive_engagement",
325
+ "desc": "The degree to which the text is likely to prompt further discussion, debate, or response from readers, based on its content and framing."
326
+ },
327
+ {
328
+ "category": "Toxicity, Controversy, and Discourse",
329
+ "dimension": "quote_worthiness",
330
+ "desc": "The degree to which the text contains memorable, quotable lines or statements likely to be shared, screenshotted, or repeated by readers."
331
+ },
332
+ {
333
+ "category": "Toxicity, Controversy, and Discourse",
334
+ "dimension": "reader_alignment",
335
+ "desc": "The degree to which the text is crafted to make readers who already agree feel validated and aligned with the author's viewpoint, reinforcing shared identity or belief."
336
+ },
337
+ {
338
+ "category": "Toxicity, Controversy, and Discourse",
339
+ "dimension": "controversy_potential",
340
+ "desc": "The degree to which the text's content or framing is likely to provoke disagreement, backlash, or polarized reactions among readers."
341
+ },
342
+ {
343
+ "category": "Toxicity, Controversy, and Discourse",
344
+ "dimension": "toxicity_score",
345
+ "desc": "The degree to which the text contains rude, disrespectful, hostile, or unreasonable content likely to make readers want to disengage from a discussion."
346
+ },
347
+ {
348
+ "category": "Toxicity, Controversy, and Discourse",
349
+ "dimension": "insult_intensity",
350
+ "desc": "The degree to which the text contains direct insults, name-calling, or demeaning language targeted at a person or group."
351
+ },
352
+ {
353
+ "category": "Factuality, Misinformation, and Credibility",
354
+ "dimension": "truth_alignment_score",
355
+ "desc": "The degree to which the claims in the text align with verifiable, established facts, based solely on internal consistency and plausibility cues within the text (not external fact-checking)."
356
+ },
357
+ {
358
+ "category": "Factuality, Misinformation, and Credibility",
359
+ "dimension": "sensationalism_index",
360
+ "desc": "The degree to which the text exaggerates, dramatizes, or sensationalizes information to provoke a strong emotional reaction rather than presenting it in a measured way."
361
+ },
362
+ {
363
+ "category": "Factuality, Misinformation, and Credibility",
364
+ "dimension": "rumor_stance",
365
+ "desc": "The degree to which the text presents unverified claims or rumors as though they were confirmed fact, rather than appropriately flagging them as uncertain or unconfirmed."
366
+ },
367
+ {
368
+ "category": "Factuality, Misinformation, and Credibility",
369
+ "dimension": "source_attribution_clarity",
370
+ "desc": "The degree to which the text clearly cites, names, or attributes its claims to identifiable sources, versus making unsourced or vaguely sourced assertions."
371
+ },
372
+ {
373
+ "category": "Factuality, Misinformation, and Credibility",
374
+ "dimension": "clickbait_potential",
375
+ "desc": "The degree to which the text (especially a headline or opening) is designed to lure engagement through curiosity gaps, exaggeration, or withheld information rather than straightforward description."
376
+ },
377
+ {
378
+ "category": "Factuality, Misinformation, and Credibility",
379
+ "dimension": "fabrication_likelihood",
380
+ "desc": "The likelihood, based on internal textual cues (implausible specifics, inconsistent details, unverifiable claims presented as fact), that content in the text has been invented or fabricated."
381
+ },
382
+ {
383
+ "category": "Persuasion and Propaganda",
384
+ "dimension": "pathos_appeal",
385
+ "desc": "The degree to which the text attempts to persuade by appealing to readers' emotions (fear, hope, anger, sympathy) rather than logic or credibility."
386
+ },
387
+ {
388
+ "category": "Persuasion and Propaganda",
389
+ "dimension": "ethos_appeal",
390
+ "desc": "The degree to which the text attempts to persuade by establishing the author's or a source's credibility, expertise, authority, or trustworthiness."
391
+ },
392
+ {
393
+ "category": "Persuasion and Propaganda",
394
+ "dimension": "loaded_language_density",
395
+ "desc": "The frequency and prominence of emotionally charged or connotation-laden words and phrases used to influence opinion beyond their literal meaning."
396
+ },
397
+ {
398
+ "category": "Persuasion and Propaganda",
399
+ "dimension": "fear_mongering_index",
400
+ "desc": "The degree to which the text deliberately amplifies fear, danger, or threat—often disproportionately to actual risk—to influence attitudes or behavior."
401
+ },
402
+ {
403
+ "category": "Persuasion and Propaganda",
404
+ "dimension": "glittering_generality_score",
405
+ "desc": "The degree to which the text uses vague, emotionally appealing but substance-light phrases (e.g., 'freedom', 'the right choice') to generate positive associations without specific justification."
406
+ },
407
+ {
408
+ "category": "Persuasion and Propaganda",
409
+ "dimension": "logical_fallacy_presence",
410
+ "desc": "The degree to which the text's argumentation relies on identifiable logical fallacies (e.g., ad hominem, straw man, false dilemma) rather than sound reasoning."
411
+ },
412
+ {
413
+ "category": "Persuasion and Propaganda",
414
+ "dimension": "call_to_action_strength",
415
+ "desc": "The degree to which the text explicitly and forcefully urges the reader to take a specific action, versus being purely informational or descriptive."
416
+ },
417
+ {
418
+ "category": "Dialogue Acts and Conversational Intent",
419
+ "dimension": "information_seeking_intent",
420
+ "desc": "The degree to which the text is framed as a request for information, clarification, or answers, rather than as a statement, assertion, or command."
421
+ },
422
+ {
423
+ "category": "Dialogue Acts and Conversational Intent",
424
+ "dimension": "agreement_score",
425
+ "desc": "The degree to which the text expresses agreement, alignment, or endorsement with a prior statement, position, or interlocutor."
426
+ },
427
+ {
428
+ "category": "Dialogue Acts and Conversational Intent",
429
+ "dimension": "disagreement_intensity",
430
+ "desc": "The degree to which the text expresses disagreement, pushback, or contradiction toward a prior statement, position, or interlocutor."
431
+ },
432
+ {
433
+ "category": "Dialogue Acts and Conversational Intent",
434
+ "dimension": "directive_intent_score",
435
+ "desc": "The degree to which the text issues instructions, commands, requests, or directives intended to get the reader/listener to do something."
436
+ },
437
+ {
438
+ "category": "Dialogue Acts and Conversational Intent",
439
+ "dimension": "commissive_intent",
440
+ "desc": "The degree to which the text commits the speaker/author to a future action, promise, or obligation (e.g., 'I will...', 'we promise to...')."
441
+ },
442
+ {
443
+ "category": "Dialogue Acts and Conversational Intent",
444
+ "dimension": "acknowledgment_presence",
445
+ "desc": "The degree to which the text contains explicit acknowledgment, recognition, or confirmation of a prior message, point, or contribution from another party."
446
+ },
447
+ {
448
+ "category": "Empathy and Psychological Signals",
449
+ "dimension": "reflection_trigger_score",
450
+ "desc": "The degree to which the text is likely to prompt the reader toward self-reflection, introspection, or reconsideration of their own views or experiences."
451
+ },
452
+ {
453
+ "category": "Empathy and Psychological Signals",
454
+ "dimension": "cognitive_empathy_score",
455
+ "desc": "The degree to which the text demonstrates understanding of another person's perspective, mental state, or reasoning, without necessarily sharing their emotion."
456
+ },
457
+ {
458
+ "category": "Empathy and Psychological Signals",
459
+ "dimension": "emotional_support_intensity",
460
+ "desc": "The degree to which the text offers comfort, reassurance, encouragement, or emotional care to another person."
461
+ },
462
+ {
463
+ "category": "Empathy and Psychological Signals",
464
+ "dimension": "distress_indicator",
465
+ "desc": "The degree to which the text signals that its author is experiencing psychological distress, such as anxiety, hopelessness, overwhelm, or crisis."
466
+ },
467
+ {
468
+ "category": "Empathy and Psychological Signals",
469
+ "dimension": "validation_score",
470
+ "desc": "The degree to which the text explicitly validates or affirms another person's feelings, experiences, or point of view as legitimate and understandable."
471
+ },
472
+ {
473
+ "category": "Empathy and Psychological Signals",
474
+ "dimension": "self_disclosure_depth",
475
+ "desc": "The degree to which the author reveals personal, private, or vulnerable information about themselves, versus keeping the text impersonal or guarded."
476
+ },
477
+ {
478
+ "category": "Humor and Wordplay",
479
+ "dimension": "humor_intensity",
480
+ "desc": "The overall degree to which the text is intended to be, or comes across as, funny, amusing, or comedic, regardless of the specific technique used."
481
+ },
482
+ {
483
+ "category": "Humor and Wordplay",
484
+ "dimension": "pun_presence_score",
485
+ "desc": "The degree to which the text uses puns or plays on words that exploit multiple meanings or similar-sounding words for comedic effect."
486
+ },
487
+ {
488
+ "category": "Humor and Wordplay",
489
+ "dimension": "absurdity_index",
490
+ "desc": "The degree to which the text relies on nonsensical, surreal, or absurd scenarios/statements for its comedic or rhetorical effect."
491
+ },
492
+ {
493
+ "category": "Humor and Wordplay",
494
+ "dimension": "self_deprecation_level",
495
+ "desc": "The degree to which the author uses humor that mocks, downplays, or pokes fun at themselves rather than at others or external subjects."
496
+ },
497
+ {
498
+ "category": "Humor and Wordplay",
499
+ "dimension": "dark_humor_probability",
500
+ "desc": "The likelihood that the text employs dark or morbid humor — finding comedy in taboo, distressing, or grim subject matter such as death, illness, or tragedy."
501
+ },
502
+ {
503
+ "category": "Political Ideology and Populism",
504
+ "dimension": "individualism_collectivism",
505
+ "desc": "The value orientation of the text, ranging from strongly collectivist (group-focused) to strongly individualistic (self-focused), with a neutral or balanced midpoint."
506
+ },
507
+ {
508
+ "category": "Emotion and Affect",
509
+ "dimension": "emotional_intensity",
510
+ "desc": "The strength of emotional expression in the text, from emotionally neutral and factual to highly passionate, dramatic, or affectively charged language."
511
+ },
512
+ {
513
+ "category": "Persuasion and Propaganda",
514
+ "dimension": "threat_and_fear_appeal",
515
+ "desc": "The extent to which the text uses danger, threats, risks, or severe negative consequences to influence readers, from no fear framing to strong fear-based messaging."
516
+ },
517
+ {
518
+ "category": "Sentiment and Polarity",
519
+ "dimension": "emotional_valence",
520
+ "desc": "The overall emotional tone of the text, ranging from clearly negative emotion to clearly positive emotion, with a neutral midpoint when emotional content is absent."
521
+ },
522
+ {
523
+ "category": "Argumentation and Epistemology",
524
+ "dimension": "scientific_thinking",
525
+ "desc": "The extent to which the text reflects scientific thinking, such as empirical evidence, experimentation, and objective description, from no scientific content to strongly scientific."
526
+ },
527
+ {
528
+ "category": "Argumentation and Epistemology",
529
+ "dimension": "rational_thinking",
530
+ "desc": "The extent to which the text reflects rational or logical reasoning, such as structured arguments, cause-effect reasoning, and an analytical tone."
531
+ },
532
+ {
533
+ "category": "Factuality, Misinformation, and Credibility",
534
+ "dimension": "evidence_based_reasoning",
535
+ "desc": "The extent to which claims rely on evidence, data, analysis, and logical justification, from unsupported opinions and anecdotes to rigorous, evidence-backed argumentation."
536
+ },
537
+ {
538
+ "category": "Sentiment and Polarity",
539
+ "dimension": "optimistic_outlook",
540
+ "desc": "How positively the text views future outcomes, from pessimistic and hopeless to highly hopeful and focused on progress and improvement."
541
+ },
542
+ {
543
+ "category": "Political Ideology and Populism",
544
+ "dimension": "social_transformation",
545
+ "desc": "The degree of support for changing existing social, political, or economic systems, from strongly preserving current structures to advocating reform or systemic change."
546
+ },
547
+ {
548
+ "category": "Political Ideology and Populism",
549
+ "dimension": "hierarchical_worldview",
550
+ "desc": "How strongly the text accepts or legitimizes unequal social power structures, from strongly egalitarian to endorsing hierarchy, authority, and rank as natural or desirable."
551
+ },
552
+ {
553
+ "category": "Political Ideology and Populism",
554
+ "dimension": "tradition_preservation",
555
+ "desc": "The emphasis placed on maintaining traditions, customs, and established practices, from rejecting or replacing them to strongly preserving cultural heritage and norms."
556
+ },
557
+ {
558
+ "category": "Argumentation and Epistemology",
559
+ "dimension": "moral_certainty",
560
+ "desc": "Whether moral judgments are presented as fixed truths or as complex and contextual, from nuanced and ambiguous to definitive claims about right and wrong."
561
+ },
562
+ {
563
+ "category": "Empathy and Psychological Signals",
564
+ "dimension": "empathetic_perspective",
565
+ "desc": "The extent to which the text explores the emotions, experiences, and viewpoints of others, from little consideration to deep empathy and multiple human perspectives."
566
+ },
567
+ {
568
+ "category": "Readability, Cohesion, and Cognitive Load",
569
+ "dimension": "conceptual_complexity",
570
+ "desc": "The complexity and abstraction of the ideas presented, from simple, concrete concepts to highly theoretical, multi-layered ideas with sophisticated reasoning."
571
+ },
572
+ {
573
+ "category": "Persuasion and Propaganda",
574
+ "dimension": "persuasive_intent",
575
+ "desc": "How strongly the text attempts to influence beliefs, attitudes, or behaviors, from purely descriptive and informational to openly advocative or ideological."
576
+ },
577
+ {
578
+ "category": "Social Media and Cultural Signals",
579
+ "dimension": "gender_orientation",
580
+ "desc": "The gender orientation of the text's language and framing, ranging from strongly feminine to strongly masculine, with a neutral or balanced midpoint."
581
+ },
582
+ {
583
+ "category": "Social Media and Cultural Signals",
584
+ "dimension": "feminine_home_work_association",
585
+ "desc": "How feminine characters are associated with context, from strongly tied to home and domestic settings to strongly tied to work and professional settings."
586
+ },
587
+ {
588
+ "category": "Social Media and Cultural Signals",
589
+ "dimension": "masculine_home_work_association",
590
+ "desc": "How masculine characters are associated with context, from strongly tied to home and domestic settings to strongly tied to work and professional settings."
591
+ }
592
+ ]
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
+ """