| Member Profile : Tore Opsahl |
|
 | Contact Information | Address: -Map Me-
Tore Opsahl
Imperial College London, Imperial Collge Business School
South Kensington
London, -, United Kingdom SW7 2AZ
Phone : +44 20 7594 3035
Fax : +44 20 7823 7685
E-mail : tore@opsahl.co.uk
Website : http://toreopsahl.com
|
| Bibliographic Information |
Seierstad, C. & Opsahl, T. (2011). For the few not the many? The effects of affirmative action on presence, prominence, and social capital of female directors in Norway. Scandinavian Journal of Management, 27 (1).
Opsahl, T. & Hogan, B. (2010). Growth mechanisms in continuously-observed networks: Communication in a Facebook-like community. arXiv:1010.2141.
Opsahl, T., Agneessens, F. & Skvoretz, J. (2010). Node centrality in weighted networks: Generalizing degree and shortest paths. Social Networks, 32 (3), 245-251.
Opsahl, T. (2010). Triadic closure in two-mode networks: Redefining the global and local clustering coefficients. arXiv, 1006.0887.
Opsahl, T. & Panzarasa, P. (2009). Clustering in weighted networks. Social Networks, 31 (2), 155-163.
Panzarasa, P., Opsahl, T. & Carley, K. (2009). Patterns and Dynamics of Users’ Behaviour and Interaction: Network Analysis of an Online Community. Journal of the American Society for Information Science and Technology, 60 (5), 911-932.
Opsahl, T., Colizza, V., Panzarasa, P. & Ramasco, J. (2008). Prominence and control: The weighted rich-club effect. Physical Review Letters, 101 (168702).
|
| | | Software & Data | | Active Calendar Listings |
| tnet: Analysis of weighted networks, two-mode networks, and longitudinal networks | (Software) |
tnet is a package written in R to serve three purposes:
1. Calculate social network measures on weighted networks
Not everyone is the same. Some people are close to us, whereas others are just acquaintances. Few network measures, and fewer network analysis programmes, can deal with datasets where the ties are differentiated by weights. By removing the weights of relations, we are removing a lot of the richness within the dataset. This means that we are limiting the weight analysis to sensitivity analyses, which are difficult to interpret. A close friendship is not the same as an acquaintance.
2. Calculate social network measures on two-mode networks (also known as affiliation or bipartite networks)
Most forms of interaction occur through mediums, such as meetings, projects, forums, etc. By simply joining two people if they have interacted with the same medium, we greatly reduce the information available to analyse. For example, the clustering coefficient on a one-mode projection of a two-mode network is meaningless as triangles are formed automatically when three or more people interact with the same medium. To remove some of the biases that might invalidate the analysis, a new set of measures directed at analysing two-mode networks directly (and a software were these measures are implemented) are needed.
3. Detect underlying principles that guide tie formation in networks with time-stamped ties (from version 3)
Network analysis is often based on static networks. In these networks there are issues of dependence as everything depends on everything. Therefore it is difficult to say why certain ties are created and others are not. In networks where the exact sequence of ties is know, the endogeneity issue can be dealt with. This type of data is generally from online communities, email networks, and telephone networks (if your dataset is not like this, but collected in waves, try Siena).
tnet is available on CRAN. Please send feedback about the package if you have any issues or suggestions. |
| Interlocking directorate in Norway (monthly since May 2002) | (Data) |
| For a recent paper on boards and gender, I downloaded information on the composition of all Norwegian public limited companies (these companies must have 40% representation of each gender on their boards of directors). The monthly networks are available in UCINET format through the paper’s supporting website (www.boardsandgender.com). The site contains both the two-mode network files and projected one-mode network files for each month since May 2002 as well as names and directors' gender. |
| Weighted network datasets | (Data) |
| A collection of weighted network datasets and two-mode datasets that I have collected or were available on the Internet, including Freeman's three EIES networks. They are available in both tnet and UCINET dl-formats. Each dataset comes with a description and suggested reference, and some with attribute files. The weighted networks comes with the definition of the weights. |
|
|
|
| | | Network Graduate Programs | | Network Courses |
|
|
|
| | | Jobs Posted | | Sunbelt Submissions |
|
| Sunbelt XXX - June 29 to July 04, 2010 - Riva del Garda Fierecongressi | | Abstract : Triadic closure in two-mode networks: Redefining the global and local clustering coefficients |
| Many network dataset are by definition two-mode networks. Yet, few network measures can directly be applied to them. Therefore, two-mode networks are often projected onto one-mode networks by selecting a node set and linking two nodes if they were connected to common nodes in the two-mode network. This process has a major impact on the level of clustering in the network. If three or more nodes are connected to a common node in the two-mode network, the nodes form a fully-connected clique consisting of one or more triangles in the one-mode projection. A number of modeling issues is associated with this procedure. For example, a one-mode projection of a random two-mode network will have a higher clustering coefficient than its randomly expected value. This paper represents an attempt to overcome these issues by redefining both the global and local clustering coefficients so that they can be calculated directly on the two-mode structure. I illustrate the benefits of such an approach by applying it to two-mode networks from four different domains: event attendance, scientific collaboration, interlocking directorates, and online forums. |
| Workshop : tnet: Software for Analysis of Weighted, Two-mode, and Longitudinal networks |
tnet is a package written in R to serve three purposes:
1. Calculate social network measures on weighted datasets
Not everyone is the same. Some people are close to us, whereas others are just acquaintances. Few network measures, and fewer network analysis programmes, can deal with datasets where the ties are differentiated by weights. By removing the weights of relations, we are removing a lot of the richness within the dataset. This means that we are limiting the weight analysis to sensitivity analyses, which are difficult to interpret. A close friendship is not the same as an acquaintance.
2. Calculate social network measures on two-mode
Most forms of interaction occur through mediums, such as meetings, projects, forums, etc. By simply joining two people if they have interacted with the same medium, we greatly reduce the information available to analyse. For example, the clustering coefficient on a one-mode projection of a two-mode network is meaningless as triangles are formed automatically when three or more people interact with the same medium. To remove some of the biases that might invalidate the analysis, a new set of measures directed at analysing two-mode networks directly (and a software were these measures are implemented) are needed.
3. Detect underlying principles that guide tie formation in datasets with time-stamped ties
Network analysis is often based on static networks. In these networks there are issues of dependence as everything depends on everything. Therefore it is difficult to say why certain ties are created and others are not. In networks where the exact sequence of ties is known, the endogeneity issue can be dealt with. This type of data is generally from online communities, email networks, and telephone networks (if your dataset is not like this, but collected in waves, try Siena). |
| Sunbelt XXXI - February 08 to February 13, 2011 - Trade Winds Beach Resort
http://www.tradewindsresort.com/
St. Pete Beach | | Abstract : Revisiting small-world networks: Is the world small? |
| Small-world networks have been found to exist in abundance. Most studies focus on observing an average distance among nodes that is comparable to one found in corresponding classical random networks. Two issues are worth considering when comparing observed properties to random ones: (1) what is a corresponding random network, and (2) how close should an observed value be to the randomly expected one to be deemed comparable. In this presentation, I will show how various types of randomisations impact on the randomly expected value of both the average distance and a second small-world property, clustering, in a range of domains, and how to determine whether the two properties are statistically non-significantly different from and significantly higher than, respectively, the randomly expected values. The results demonstrate that small-world networks are far from as abundant as previously thought, and only a few networks satisfy both properties. |
| Workshop : tnet: Software for Analysis of Weighted, Two-mode, and Longitudinal networks |
This is the second time that I will hold a workshop on the R-package tnet. Based on feedback from Riva del Garda, this workshop will focus more on how the measures are defined.
Like last year, the workshop will be on weighted and two-mode networks:
1. Calculate social network measures on weighted datasets
Not everyone is the same. Some people are close to us, whereas others are just acquaintances. Few network measures, and fewer network analysis programmes, can deal with datasets where the ties are differentiated by weights. By removing the weights of relations, we are removing a lot of the richness within the dataset. This means that we are limiting the weight analysis to sensitivity analyses, which are difficult to interpret. A close friendship is not the same as an acquaintance.
2. Calculate social network measures on two-mode
Most forms of interaction occur through mediums, such as meetings, projects, forums, etc. By simply joining two people if they have interacted with the same medium, we greatly reduce the information available to analyse. For example, the clustering coefficient on a one-mode projection of a two-mode network is meaningless as triangles are formed automatically when three or more people interact with the same medium. To remove some of the biases that might invalidate the analysis, a new set of measures directed at analysing two-mode networks directly (and a software were these measures are implemented) are needed. |
| Sunbelt XXXII - March 12 to March 18, 2012 - Crowne Plaza Hotel
https://resweb.passkey.com/go/INSNASunbelt
Redondo Beach | | Workshop : tnet: Software for Analysis of Weighted, Two-mode, and Longitudinal Networks |
| everything else. Thus, it is difficult to say why certain ties are created and others are not. In networks where the exact sequence of ties is known, the endogeneity issue can be dealt with. This type of data is generally from online communities, email networks, telephone networks, and network data collected using RFID tags (if your dataset is not like this, but collected in waves, try Siena). |
|
|
| |
|
