Skip to content
/ mdic Public

Code and data for the paper: "Message Distortion in Information Cascades" (TheWebConf2019)

Notifications You must be signed in to change notification settings

epfl-dlab/mdic

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Message Distortion in Information Cascades

This repository contains the data and code of the paper "Message Distortion in Information Cascades" (which you can read clicking here):

@inproceedings{horta_ribeiro_message_2019,
author={Ribeiro, Manoel Horta and Gligori\'c, Kristina and West, Robert},
title={Message Distortion in Information Cascades},
booktitle={Proceedings of the 2019 World Wide Web Conference},
year={2019},
}

Check out the accompanying website which allows you to visualize the data.

Data

You may find the data in the data folder ./data (duh).

We make the data available in two formats: a .csv and .graphml. The latter is the format used in the analysis of the data (for convenience).

Field Null in Root Description
node_id No Unique identifier of the node.
level No Summarization level: (0: original abstract, 1: ~1024 chars, 2: ~512 chars, 3: ~256 chars, 4: ~124 chars, 5: ~64 chars)
branch No Source note used in this information cascade.
question No node_id of the text used as reference for the summarization, in case of root nodes, it is the same as node_id
Topic No Topic of the paper summarized (breast, cardio, immunization, diet)
Answer No Original abstract in the case of the root, summary otherwise.
Age Yes Age range of the worker which summarized the paper (18-24/24-39/40-60/60+)
Education Yes Education level of the worker which summarized the paper (Some High School, High School, Some College, College)
Gender Yes Gender of the worker which summarized the paper (male, female)
Qualification Yes Performance on qualification test (float, 0-1)
WorkerId Yes Unique worker identifier as provided by amazon mt
Doggos_crowd No Dictionary containing the values for facts in each category. {"Coarse":{"Coarse_category1":["Val1", "Val2", ...] ...}, "Fine":{"Fine_category1":["Val1"} ...}
Doggos_text No Dictionary containing the text for facts of each sub category. Null in non-root. {"Fine_category1": "Text1", "Fine_category2": "Text2", ... }
Tagging No In the csv files, this is a Dictionary, similar to doggos crowd, containing the keyphrases associated with each subcategory. {"Coarse": {"keyword1": {"Course_category1", ...}, ... }, "Fine": {"keyword1": {"Fine_category1", ...}, ... }. For the graphml files, this is actually a python object with this dictionary, and a bunch of helper functions to calculate the difference in keywords across hops.

Code

All the analysis performed may be found in the analyses.ipynb notebook.

To install all requirements simply run

pip install -r requirements.txt 

About

Code and data for the paper: "Message Distortion in Information Cascades" (TheWebConf2019)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages