Skip to main content

JoSS: Journal of Social Structure

Popularity Report

Total Popularity Score: 0
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...

Rank

Bookmark History

Saved by 26 people (-4 private), first by anonymouse user on 2007-06-27


Public Comment

on 2007-07-04 by christyinsdesign

Research with interesting methods of visualizing networks and response patterns in online discussion groups. This paper is focused mostly on people who play the role of "answer person" in these groups, the people who answer lots of questions for others but don't ask too many questions of their own.

Public Sticky notes

Visualizing the Signatures of Social Roles in Online Discussion Groups

Highlighted by djiezes

Highlighted by yassinelhassani

Answer people are individuals whose dominant behavior is to respond to questions posed by other users

Highlighted by yassinelhassani

answer people

Highlighted by yassinelhassani

are disproportionately tied to relative isolates, have few intense ties and have few triangles in their local networks

Highlighted by yassinelhassani

role behavior

Highlighted by yassinelhassani

OLS regression

Highlighted by yassinelhassani

identification of role behavior in online discussion settings and consider how the development of a taxonomy of author types could be extended to a taxonomy of newsgroups in particular and discussion systems in general.

Highlighted by yassinelhassani

The central premise of this paper is that we should be able to recognize the roles that people play by measuring behavioral and structural “signatures” of their participation.

Highlighted by sshein

People contribute to these conversations for a variety of reasons: for debate, to express appreciation or affiliation, to build a sense of community, to provide and receive social support, to collect information, and to provide answers to questions. When people systematically participate in a particular manner, they can be seen as playing a social role.

Highlighted by yassinelhassani

role of a fan

Highlighted by yassinelhassani

he answer person is a well recognized social role in online discussion spaces (Golder 2003; Viegas and Smith 2004; Turner, Fisher, Smith, and Welser 2005). Substantively, they are important because answer people collectively donate vast amounts of valuable advice to those who ask questions, resulting in the creation of valuable online resources.

Highlighted by sshein

we should be able to recognize the roles that people play by measuring behavioral and structural “signatures”

Highlighted by yassinelhassani

Golder 2003; Viegas and Smith 2004; Turner, Fisher, Smith, and Welser 2005

Highlighted by yassinelhassani

the creation of valuable online resources

Highlighted by yassinelhassani

Wenger, McDermott, and Snyder 2002

Highlighted by yassinelhassani

We identify a structural signature for answer people with three features based on the local network attributes and the types of threads to which answer people tend to contribute. First, answer people tend to be disproportionately connected to alters with low degree. That is, they reply to relative isolates, authors who themselves answer few, if any, others. Second, their local (degree one) networks tend to have small proportions of three-cycles (i.e. their neighbors are not neighbors of each other) and they seldom send multiple messages to the same recipient (few intense ties). Finally, answer people tend to reply to discussion threads initiated by others and typically only contribute one or two messages per thread.

Highlighted by sshein

altruism

Highlighted by yassinelhassani

contributing to public goods for social goods like status

Highlighted by yassinelhassani

Highlighted by sshein

we use visualizations of networks of reply connections and the behavioral histories of contributions to develop and extend intuitions about distinctive structural features of the answer person role

Highlighted by yassinelhassani

First, answer people tend to be disproportionately connected to alters with low degree. That is, they reply to relative isolates, authors who themselves answer few, if any, others. Second, their local (degree one) networks tend to have small proportions of three-cycles (i.e. their neighbors are not neighbors of each other) and they seldom send multiple messages to the same recipient (few intense ties). Finally, answer people tend to reply to discussion threads initiated by others and typically only contribute one or two messages per thread.

Highlighted by yassinelhassani

Social roles are a foundational concept in social analysis, but one that lacks a clear definition and method for their identification based on objective measures

Highlighted by yassinelhassani

finding roles in online data will be helpful for users and organizers of online forums who may be able to use these techniques to build reputation systems which identify helpful or deleterious users.

Highlighted by yassinelhassani

Developing methods for finding roles in online data is especially important because such data are increasingly available and often has multiple dimensions that can be leveraged simultaneously, allowing relative accuracy of the different methods to be compared.

Highlighted by clakesnapster

There are many important social roles in online discussion groups: local experts, answer people, conversationalists, fans, discussion artists, flame warriors, and trolls

Highlighted by yassinelhassani

on 2008-12-01 by yassinelhassani

Les rôles types identifiés: - Local Expert - Answer People - Conversationalists - Fans - Discussion Artists - Flame Warrior - Trolls (Personnes qui aiment lancer des polémiques)

online settings are ideal for studying roles because they allow researchers to simultaneously bring network structure, behavioral patterns, and the meaning of interactions (via content analysis) to bear on the task of accurately identifying roles.

Highlighted by clakesnapster

The best answer people will provide prompt, accurate, and thorough help. Participants who enact this role are an important source of the value found in online discussion groups. They offer this help to potential strangers without direct compensation or expectation of reciprocity and, often, without thanks.

Highlighted by yassinelhassani

on 2008-12-01 by yassinelhassani

Rappel sur l'importance du rôle d'Answer People et de leur motivations

important social roles in online discussion groups: local experts, answer people, conversationalists, fans, discussion artists, flame warriors, and trolls

Highlighted by clakesnapster

structural signature' to refer to distinctive attributes that define types of networks

Highlighted by yassinelhassani

Networks and Structural Signatures of Social Roles. In general, network studies of roles are premised on the notion that structural similarity indicates a class of similar actors which may correspond to social roles (White, Boorman and Breiger 1976; Winship 1988). In particular, the family of equivalence methods is often used to break a population of actors into classes based on graph-wide measures, and relationships among these classes are used to identify roles (Wasserman and Faust 1994). However, we are less concerned with assigning all actors to different classes than we are in identifying general structural features that are associated with one particular role. Therefore we use degree one ego-centric network data and visualization to identify structural attributes associated with the answer person role.

A handful of network studies provide helpful models for our techniques. Mizruchi (1993) compared the relative ability of measures of cohesion and structural equivalence to predict a corporation’s political donation activity. Although the explanatory power of the models was limited, this study is important because it tested the adequacy of the equivalencies. Ronald Burt’s inquiry into structural holes emphasizes how metrics based on local network measures can identify important types of social actors at the edge of structural holes (1992, 2004). Erickson’s work on bridge-playing teams highlighted the relevance of structural characteristics to the role of a prestigious player (1984). Our research follows these by testing structural indicators against reliable measures of role behavior like the content of their messages.

Highlighted by clakesnapster

a major research challenge is to develop the ability to identify groups with good information and identify valuable contributors without inspecting every message in a collection.

Highlighted by sshein

Haythornthwaite and Hagar

Highlighted by yassinelhassani

on 2008-12-01 by yassinelhassani

Reference

The people who earn good reputations may well be those who are skilled at boosting their reputation scores and not those who are most helpful

Highlighted by yassinelhassani

on 2008-12-01 by yassinelhassani

Les systemes de réputation basés sur la notation de commentaires sont détournés. Rrésultat : Les contributeurs qui ont une excellente réputation ne sont pas ceux qui apportent le plus d'aide et de contenu de valeur

challenge is to develop the ability to identify groups with good information and identify valuable contributors without inspecting every message in a collection.

Highlighted by yassinelhassani

on 2008-12-01 by yassinelhassani

Rappel de l'enjeu : identifier les contributeurs de valeurs sans devoir inspecter tous les messages

playful banter, story-telling, bragging, confrontation, announcement and promotion of events, products, or services.

Highlighted by yassinelhassani

on 2008-12-01 by yassinelhassani

Formes d'échanges : Playful Banter : Badinage Story-Telling Bragging : Vantardise Confrontation Promotion ou annonce d'évenement

Figure 7b:  Contributor to Kites who degree distribution is similar to the answer person pattern

Highlighted by taylorparsons

Typology of Social Roles in Usenet

Highlighted by yassinelhassani

[3] The qualitative coding of authorlines limits this study because of possibility of bias in the measurement process. We minimized this bias by coding author lines while obscuring the author’s identity and answering behavior, however some familiarity was unavoidable due to distinctiveness of extreme examples in the authorlines. Subsequent work will replace hand coding with direct measurement of variables on which the authorlines visualization is based. Unfortunately those variables are not available for these data at this time, but we are adapting the database to make this possible for future work.

Highlighted by clakesnapster