convote_mentions — US House floor debate mentions (ConVote, 2005)

Description

A directed signed network of the by-name references between the speakers of floor debates of the US House of Representatives in 2005, from the ConVote corpus of Thomas, Pang and Lee (2006). The authors downloaded the 2005 record of the House from govtrack.us, associated each page with the bill it discussed most, kept the bills that came to a vote, split the pages into speech segments, and labeled each segment by the vote (yes or no) of its speaker on the bill; speakers who abstained were discarded, and only the debates (the speeches on a bill) with at least 20% of speeches of each label were kept, 53 debates in total. In the speeches, the references by name to other members of the House who spoke in the same debate were identified, and a classifier (a support vector machine) was trained on those of the training set (39 debates) to tell whether a reference expresses agreement or disagreement with the referenced member.

The network contains the references of the 14 debates of the development and test sets (the references of the training set are not distributed), with the scores of the classifier. Each node is a speaker (a member of the House), and a directed edge from speaker A to speaker B is a reference to B in a speech of A in a debate. Pairs of speakers may be connected by several edges, and speakers may refer to themselves. The sign of an edge is the true label of the reference: agreement if both speakers voted the same way on the bill, and disagreement otherwise. Only the speakers making or receiving some reference are included (219 of the 302 speakers of these debates). The texts of the speeches, the scores of the individual speeches by the classifier of support or opposition (and the strengths of the corresponding links of the authors' graphs), and the indicators of whether a speech mentions the bill directly are not included.

The network has the following properties:

(Node) speaker_id: the index of the speaker in the corpus.
(Node) party: the party of the speaker ("D" for Democratic, "R" for Republican, "I" for independent).
(Edge) sign: +1 for a reference between speakers who voted the same way on the bill (agreement), -1 otherwise (disagreement).
(Edge) debate: the index of the bill under discussion (identifying the debate).
(Edge) source_vote and target_vote: the votes ("yes" or "no") of the referring and the referenced speaker on the bill.
(Edge) raw_score: the score of the agreement classifier (positive for references classified as agreements).
(Edge) normalized_score: the score divided by the standard deviation of the scores of the references of the debate.
(Edge) strength: the strength of the link between the speakers in the authors' graph, 2500 times the normalized score (0 for negative normalized scores).
(Edge) normalized_score_high_precision and strength_high_precision: the same, but with the score minus the average score of the debate, which keeps only the references with above-average scores in the graph.

Tags

Social Political Signed Multigraph Metadata

Added

2026-10-27

Citation

M. Thomas, B. Pang, L. Lee, "Get out the vote: Determining support or opposition from congressional floor-debate transcripts." Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing, 327-335 (2006) 10.3115/1610075.1610122 Sci-Hub

Upstream URL OK

https://www.cs.cornell.edu/home/llee/data/convote.html

Networks

Tip: hover over a column header for a description.

Column legend

$\left<k\right>$: Mean degree
$\sigma_k$: Standard deviation of the degrees
$\lambda_h$: Largest eigenvalue of the Hashimoto (non-backtracking) matrix
$\tau$: Random walk mixing time
$r$: Degree assortativity
$c$: Global clustering coefficient
$\oslash$: (Pseudo-) diameter
$S$: Relative size of the largest component
NPs: Node properties
EPs: Edge properties

Name	Nodes	Edges	$\left<k\right>$	$\sigma_k$	$\lambda_h$	$\tau$	$r$	$c$	$\oslash$	$S$	Kind	Mode	NPs	EPs	gt	GraphML	GML	csv
convote_mentions	219	764	3.49	9.04	8.31	8.05	-0.21	0.15	7	1.00	Directed	Unipartite	speaker_id party	sign debate source_vote target_vote raw_score normalized_score strength normalized_score_high_precision strength_high_precision	27 KiB	29 KiB	28 KiB	24 KiB

Ridiculograms^*

^* These are automatically generated force-directed visualizations, and can be quite meaningless for networks both big and small. They should not be taken seriously as sources of scientific insight. See here for a discussion.