Chat Log Text Analysis

Introduction

Unstructured data in the form of text encompasses a large portion of all data stored and can be sourced from surveys, Twitter posts, video transcripts, and chat logs amongst a host of other locales. This application surveys the types of analysis that can be done to bring forth insights that are hidden from more traditional qualitative approaches.

Plot Options

User Options

Video Tutorial

Data

The chat logs analyzed in this application come from an African non-profit that offers support for individuals who have under gone a traumatic experience at the hands of their regional government. These religious workers spend hours comforting victims through a remote chat-room.

Sample Chat Log Text

21/09/2015 07:01 Blessing Mureverwi Good morning and thank you
21/09/2015 07:03 Blessing Chikaka 4 For whatsoever things were written aforetime were written for our learning, that we through patience and comfort of the scriptures might have hope.
21/09/2015 07:04 Blessing Chikaka Varoma 15 ndima yemukati 4
21/09/2015 07:04 Blessing Chikaka Morning M (h)ute
21/09/2015 07:06 Blessing Chikaka 8 Owe no man any thing, but to love one another: for he that loveth another hath fulfilled the law. 9 For this, Thou shalt not commit adultery, Thou shalt not kill, Thou shalt not steal, Thou shalt not bear false witness, Thou shalt not covet; and if there be any other commandment, it is briefly comprehended in this saying, namely, Thou shalt love thy neighbour as thyself. 10 Love worketh no ill to his neighbour: therefore love is the fulfilling of the law.
21/09/2015 07:07 Blessing Mureverwi Habari za asubuhi
21/09/2015 07:09 Blessing Chikaka 9 Let love be without dissimulation. Abhor that which is evil; cleave to that which is good. 10 Be kindly affectioned one to another with brotherly love; in honour preferring one another;
21/09/2015 07:10 Blessing Mureverwi Rudo ngarurege kuva nokunyengera...powerful
21/09/2015 07:14 Blessing Chikaka 4 Rudo runomoyo murefu, runomoyo munyoro; rudo harunegodo; rudo harunamanyawi, haruzvikudzi; 5 haruiti zvisingafaniri, haruzvitsvakiri zvarwo; harutsamwi, harunezvishura; 6 harufariri zvisakarurama, asi runofarira zvokwadi; 7 runofukidza zvose, runotenda zvose, runetariro pazvose, runotsungirira pazvose. 8 Rudo harutongoperi;
21/09/2015 07:15 Blessing Chikaka 1 Saka, zvatinako kushumira uku, patakanzwirwa ngoni napo, hatineti;
21/09/2015 07:15 Blessing Chikaka Ishe akomborere mashoko ake nokusingaperi
21/09/2015 07:16 Tari Dzidza kupota uchinyora maVerse.
21/09/2015 07:17 Steven Mazivanhanga Titus 3:4-6 But when the kindness and love of God our Savior appeared, [5] he saved us, not because of righteous things we had done, but because of his mercy. He saved us through the washing of rebirth and renewal by the Holy Spirit, [6] whom he poured out on us generously through Jesus Christ our Savior, Goodmng! Be blessed Mhuterians
21/09/2015 07:17 Tari I mean where they came from
21/09/2015 07:17 Tari Asi arikunakidza aya
21/09/2015 07:17 Tari πŸ˜„
21/09/2015 07:17 Tari Guten Morgen
21/09/2015 07:19 Steven Mazivanhanga
21/09/2015 07:19 Mahovo Apa maitawo mhuteπŸ‘πŸΎπŸ‘πŸΎπŸ‘πŸΎ
21/09/2015 07:22 Blessing Chikaka 21 Ndivudzei, imi munoda kuva pasi pomurairo, hamutereri murairo here?
General statistics on the log file allow us to gain a high-level macro view of the usage of the site. Here we can characterize usage by time and frequency through visualizations that assist in elucidating their patterns. Users can quickly answer questions such as 'What day is activity the highest?' and 'How many people regularly use my site on Fridays?'

Stats


Plots

For individual analysis, features can be extracted directly from the text of individuals, analyzed, then categorized to create a profile of the user's language. For marketers, this information can be valuable for understanding the communication style of their target population, more effectively allowing them to draft the most relevant narrative possible. Within the security domain, groups of users can be categorized according to their communication patterns to identify hierarchy within networks. Feature extraction and lexicon analysis can be beneficial supplements to a robust text analysis.
Network analysis on text provides a quick understanding of whom are the influencers within a group, individuals who could later serve as efficient points of contact. By applying specialized algorithms, we can graph the network of discussions in a forum to determine not only who's being spoken to, but whom they address in response. The graph provides a unique understanding or the structure of group.