This folder contains the data used in the paper:

Stefano Menini, Federico Nanni, Simone Paolo Ponzetto and Sara Tonelli. “Topic-Based Agreement and Disagreement in US Electoral Manifestos”. In Proceedings of the 2017 Conference on Empirical Methods for Natural Language Processing.

The data consists of three resources:
- the macro-domain annotations of the U.S. electoral manifestos.
- the key-concepts extracted from each macro-domain and used for the creation of the topics.
- the annotations of agreement and disagreement on the topics created by clustering the key-concepts.

This data is licensed under a Creative Commons Attribution 4.0 International License.
https://creativecommons.org/licenses/by/4.0/

### MANIFESTOS ANNOTATIONS ###

The file “manifestos-domains.tsv” contains the Republican and Democratic manifestos from 2004 to 2012.
Each sentence in the manifestos has been automatically annotated with a macro-domain using the ClassyMan system (Zirn et al. 2016).

The columns in the file are respectively:
party - year  - sentence  - macro-domain 

The macro-domains are represented with a number as follows:
0 - other
1 - external relations
2 - freedom and democracy
3 - political system
4 - economy
5 - welfare and quality of life
6 - fabric of society
7 - social groups



### KEY-CONCEPTS LISTS ###

The folder “key-concepts_lists” contains the lists of key-concepts, divided by year and macro-domain, extracted from the manifestos with KD: Keyphrase Digger (http://dh.fbk.eu/technologies/kd).
Each key-concept in the lists is associated with its tf-idf score over the different macro-domains.



### AGREEMENT/DISAGREEMENT ANNOTATIONS ###

The file “manifestos_agreement.tsv” contains pairs of sentences about the same topic annotated as in agreement, disagreement or neutral.
Each pair contains three sentences from Republican manifestos on one side and three sentences from Democratic manifestos on the other side. 
For more details about the creation of these pairs please check the paper.

The columns in the file are respectively:
id - macro-domain - annotation (agreement/disagreement/neutral) - topic - democratic_sentence_1 - democratic_sentence_2 - democratic_sentence_3 - republican_sentence_1 - republican_sentence_2 - republican_sentence_3