ljcamargo commited on
Commit
b38a51a
·
verified ·
1 Parent(s): 9b82203

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +51 -1
README.md CHANGED
@@ -15,12 +15,13 @@ tags:
15
  - méxico
16
  pretty_name: Hasta Siempre Presidente
17
  ---
18
- Here is the markdown with the file links, using the provided URL root:
19
 
20
 
21
  # Hasta Siempre Presidente
22
  Dataset de la versiónes estenográficas de todas la conferencias "mañaneras" y actos públicos de la presidencia de Andrés Manuel López Obrador, México (2018-2024)
23
 
 
 
24
  ### Obtención
25
  Obtenido de la página oficial de la presidencia en aquel momento, [https://www.gob.mx/presidencia](https://www.gob.mx/presidencia/es/archivo/articulos) obtenidos automatizadamente, homologados, limpiados y estructurados en diálogos en jsonl para permitir el entrenamiento de modelos de lenguaje o su investigación periodística.
26
  Para la obtención de roles se realizó estadísita de los hablantes identificados para distinguir personas funcionarias de audiencia y reporteros que realizan preguntas.
@@ -73,11 +74,60 @@ Para la obtención de roles se realizó estadísita de los hablantes identificad
73
  [posts_similarities.py](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/posts_similarities.py)
74
 
75
 
 
 
 
 
 
 
 
 
 
 
 
76
 
 
 
 
 
 
 
77
 
 
78
 
 
 
79
 
 
 
 
80
 
 
 
81
 
 
 
82
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
83
 
 
15
  - méxico
16
  pretty_name: Hasta Siempre Presidente
17
  ---
 
18
 
19
 
20
  # Hasta Siempre Presidente
21
  Dataset de la versiónes estenográficas de todas la conferencias "mañaneras" y actos públicos de la presidencia de Andrés Manuel López Obrador, México (2018-2024)
22
 
23
+ (english at the bottom)
24
+
25
  ### Obtención
26
  Obtenido de la página oficial de la presidencia en aquel momento, [https://www.gob.mx/presidencia](https://www.gob.mx/presidencia/es/archivo/articulos) obtenidos automatizadamente, homologados, limpiados y estructurados en diálogos en jsonl para permitir el entrenamiento de modelos de lenguaje o su investigación periodística.
27
  Para la obtención de roles se realizó estadísita de los hablantes identificados para distinguir personas funcionarias de audiencia y reporteros que realizan preguntas.
 
74
  [posts_similarities.py](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/posts_similarities.py)
75
 
76
 
77
+ ---
78
+
79
+
80
+ # Farewell Presidente
81
+ Dataset of the stenographic versions of all "mañaneras" press conferences and public events held by the presidency of Andrés Manuel López Obrador, Mexico (2018-2024)
82
+
83
+ ### Source
84
+ Obtained from the official website of the presidency at the time, [https://www.gob.mx/presidencia](https://www.gob.mx/presidencia/es/archivo/articulos), acquired automatically, standardized, cleaned, and structured into dialogues in jsonl format to allow training language models or conducting journalistic research.
85
+ Speaker roles were identified through statistical analysis to distinguish officials from audience members and reporters asking questions.
86
+
87
+ ## Content
88
 
89
+ ### Speaker Roles:
90
+ - President: Andrés Manuel López Obrador
91
+ - System: Includes narrators, prompters, hosts, videos, and musical stings included in the stenographic version
92
+ - Others: Other officials and speakers, excluding audience members and reporters
93
+ - Interlocutors: Audience members or reporters asking questions to the President or other officials
94
+ [speaker_to_role.json](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/speaker_to_role.json)
95
 
96
+ ### All Dialogues, with all speakers in Phi3 Chat Dialogue Format
97
 
98
+ #### President, other officials, interlocutors, and system
99
+ [phi3_dialogs_allspeakers_unfiltered.jsonl](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/phi3_dialogs_allspeakers_unfiltered.jsonl)
100
 
101
+ ### Dialogues without other officials, includes interlocutors and system
102
+ #### President, interlocutors, and system
103
+ [phi3_dialogs_mainspeakers_unfiltered.jsonl](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/phi3_dialogs_mainspeakers_unfiltered.jsonl)
104
 
105
+ #### President, interlocutors (without system)
106
+ [phi3_dialogs_mainspeakers_filtered.jsonl](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/phi3_dialogs_mainspeakers_filtered.jsonl)
107
 
108
+ #### President and interlocutors (without system) grouped by question and answer pairs
109
+ [phi3_dialogs_mainspeakers_qa_pairs.jsonl](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/phi3_dialogs_mainspeakers_qa_pairs.jsonl)
110
 
111
+ #### President and interlocutors (without system) grouped by strict question order, in sequence
112
+ [phi3_dialogs_mainspeakers_qa_strict.jsonl](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/phi3_dialogs_mainspeakers_qa_strict.jsonl)
113
+
114
+ ### Raw files, not structured into dialogue
115
+
116
+ #### Dialogues with all speakers
117
+ (President, system, others, interlocutors)
118
+ [posts_processed.jsonl](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/posts_processed.jsonl)
119
+
120
+ #### Dialogues without other officials
121
+ (President, system, interlocutors)
122
+ [posts_qa_processed.jsonl](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/posts_qa_processed.jsonl)
123
+
124
+ #### Unstandardized download of the pages
125
+ [posts.jsonl](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/posts.jsonl)
126
+
127
+ ### Statistics and Standardization
128
+ #### Speaker and Content Statistics
129
+ [posts_stats.py](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/posts_stats.py)
130
+
131
+ #### Similarities between speakers (for standardization)
132
+ [posts_similarities.py](https://huggingface.co/datasets/ljcamargo/hastasiemprepresidente/blob/main/posts_similarities.py)
133