The ROBUST Project - Managing Online Business Communities

Thomas Gottron Institute for Web Science and Technologies
University of Koblenz-Landau
56070 Koblenz, Germany


Online business communities constitute rich ecosystems that enable an amorphous group of employees, customers or company partners to disseminate, exchange and curate content items, information and valuable knowledge. While the platform technologies for online business communities have been around for several decades, there still is a lack of metrics and tools for the analysis and management of the assets provided by these communities. The ROBUST project is dedicated to fill this gap. It provides methods and tools to analyse and monitor online communities for arising risks which might threaten the health or well being of a community as well as for opportunities which bear the potential to enrich the value or the effectiveness of the community. ROBUST is embedded in three real world use cases, addressing different types and needs of online business communities: (a) an internal collaboration platform for employees, (b) a setting where a company interacts with customers and business partners to provide support for their products and (c) a public domain setting of the general public discussing about companies on the Social Web. This article motivates the need for managing online communities and gives an overview of the setting of the ROBUST project. It explains how risks are monitored, detected and treated in the risk management framework developed for ROBUST and, thereby, gives a synopsis of the methods implemented to support community managers.

Online communities have become an integral part of every day life. Highly popular communities like Facebook, Twitter, YouTube and Wikipedia are used world-wide by millions of users. But also smaller and less famous communities addressing particular geographic regions, specific user groups or niche interests thrive on the Web. In all of these communities, users exchange information, interact with each other, collaborate and cultivate their social network.

Online communities in a business context have very similar aims. Employees, customers and business partners can discuss, network with other users, and curate business relevant information. The information and knowledge provided in online business communities as well as the social interaction represent a valuable asset to the companies with which they are associated. This value is also reflected by the efforts and resources invested to establish such business communities. Software licenses, wages of administrative personnel, the work time of the employees contributing actively to the community and hardware resources cause non-recurring as well as running  costs. Accordingly, such a valuable asset needs to be taken care of and managed appropriately. If users in a business community do not find the information they are looking for, if users become inactive and leave the community or if conflicts and inappropriate language lead to an unproductive use of the system, the value of the online business community decreases. The community becomes 'unhealthy' and does not function as intended. In the worst case the community 'dies'', i.e. no productive use can be observed any longer.

This calls for the role of a community manager to oversee and care for the business community in order to protect and develop its value. The job of the community manager cannot be done manually, but requires an automatic community monitoring and risk management framework. But, while the platform technologies for online business communities have been around for several decades, there still is a lack of metrics and tools for the analysis and management of these communities. Such metrics and managements tools are a pre-requisite to analyse the state of health of a community, to identify risks and threats and to being able to react to these risks on time, well informed  and with the appropriate counter measures.

The ROBUST project is developing metrics, methods, and tools to fill this gap. Its aim is to develop a framework for monitoring online communities, to define and track risks and opportunities, and to provide a tool box of treatment plans. This article gives a high-level overview of the project and its approach to support community managers in their task.

The ROBUST project is embedded in three prototypic real-world use case scenarios: (a) an internal collaboration platform for employees, (b) a setting where a company interacts with customers and business partners to provide support for their products and (c) a public domain setting of the general public discussing about companies on the Social Web.

Employee Use Case. Many companies use intranet installations of online communities to organize the information  available internally to the employees.
This can be done using simple, off-the-shelf tools, such as Wikis, but also by using full-fledge community platforms, such as IBM Connections.
In the latter case, users are not restricted to only contribute and curate content items, but are provided with a wide range of communication channels for interacting with each other. They can cultivate their personal network of colleagues and internal contacts. More even, it is possible to establish sub-communities around certain topics, projects or other business relevant entities.
The flow of information, the constructive use of the platform and productive collaboration are typical aims of establishing such an internal business community of employees.

Business Partner Use Case. The interaction with business partners can be be implemented in an online community as well. Product support and consultancy can be done via question and answer forums and discussion boards. This setup has several benefits for all stakeholders. Using public and multilateral rather than a private and bilateral communication allows for a better documentation and re-use of information and knowledge shared and created in support processes. Furthermore, its is even possible that customers can help each other directly, based on their own experience and expertise. Such a behaviour can be motivated and rewarded with a status point system implemented in the online community. Such a policy is of particular interest when third party consultants can use the community platform to build reputation, demonstrate their expertise and eventually conclude new contracts from this context. Overall, the aim of establishing an online community with business partners can be a faster and better response to customer requests as well as a good integration with third party business partners.

Public Domain Use Case. Also the public Web with its multitude of online communities is a field of interaction for companies. Interacting with the general public, reaching out to questions, discussion and opinions uttered about the company and its products are of interest in business scenarios. In this setting, the focus is much more on monitoring online communities than on reaction to risks. After all, the possibilities for reactions and interventions are very limited on public platforms. However, already the availability of intelligence can create awareness of risks and might allow for adequate reactions.

A fundamental ingredient for business community management is tracking and monitoring. This means to constantly survey a community and capture the essential data and features. This is not a trivial task, as even moderate-sized communities generate a huge amount of data in the form of content items, social relations, interaction profiles, access logs, etc. The ROBUST project addresses the challenge of managing large-scale community data with a mix of technologies. The two main methods are stream processing of live community data [1] and the batch computation of huge data collections in a distributed and parallel fashion [2]. This context serves to compute first order features, such as sentiments, topics, clustering coefficients, network structures and interaction data.

Once the raw community data has been filtered and aggregated and essential features have been computed, the next step is to derive second order features and metrics.
This comprises, for instance, the computation of role and behaviour composition models, describing which patterns can be observed on the interactions of the users [3]. Another example are insights in the quality of contents provided and discussed in online communities [4] or in the diffusion of discussion topics in the community [5]. A third object of interest are network structures, \eg which relations do appear between a user and contents or other users, how are these connections distributed, how are they generated and how do they disappear over the course of time [6].

Together, the first and second order features serve for describing the state of a community, to capture relevant information about its health and allow for the subsequent definition of risks and opportunities.

Monitoring and keeping  track of the development of a community allows for the definition of risks and opportunities. A risk corresponds to an event occurring in the future with a certain probability and this event having a negative impact on the business community. For instance, such a risk can be that important users stop contributing contents to the community or that support requests are not answered within a pre-defined time frame. The gravity of a risk depends on the probability that the related event occurs and how serious are the negative consequences for the community. Opportunities are - on a technical level - not very different. They simply relate to future events with a positive effect.

Accordingly, the task of risk (and opportunity) detection in online communities relates to estimating the likelihood of events and their impact on the community. Given the different types of business communities and the potentially different aims and objectives of the community owners, ROBUST takes the approach of letting the users define events and their positive or negative impact on a community.

The risk detection framework is responsible for estimating the probability of these events occurring. To this end, compartment models are trained on historic data and are used to forecast the future development of an online community. Predictor services track the development of a community and evaluate the chances of events to occur which are associated with a risk or an opportunity. If the likelihood of such a event together with the gravity of the impact reaches a certain threshold, the risk detection framework raises an alarm and notifies the community manager.

Once the community manager is  warned  about an impeding risk, the ROBUST framework triggers a risk treatment process. A risk treatment process is composed of options for how to react to the detected risk as well as tools to analyse how suitable the proposed treatments are. For a concrete risk, theses options constitute a risk treatment plan which guides the community manager and proposes actions that can be taken.

A typical first action is to investigate the origin of the risk and have the community manager take a look at the data and the community itself. Visualization techniques of network structures, topic distributions, the role composition in a community, etc. allow for a concrete evaluation of the situation. The tools shown in Figure 1, for instance, allow for analysing in parallel the temporal evolution of global community metrics as well as the network structures between users and contents. Depending on the type of risk, different visualizations will be appropriate to implement the right visual analytics. Once a human user has assessed the risk, he or she might better be able to judge if and which next actions to take.

Fig. 1. Visualization tools allow for the comparison of the temporal evolution of aggregated community metrics with the dynamics of network structures.

Simulation tools provide further support in the decision process of which reactions to adopt. They allow for evaluating what-if scenarios, in which the community manager can play around with settings for different reaction options. For instance, a sensible option to counter a sudden increase of spam in a discussion forum can be to increase the number of discussion moderators [7]. The question in this context obviously is, how many moderators are necessary to effectively reduce the amount of spam. Using simulation tools, the community manager can investigate the effects achieved for concrete settings or even alternative treatment options.

The analytical components for community data, the risk predictor services and treatment plans including the simulation functions are fully implemented and integrated in a first prototype. The next step of the ROBUST project is to provide a public demonstrator of this prototype, operating on a publicly accessible online community. The prototype will serve as a showcase for how management of an online business community could look like in the future. Users on the Web will be able to observe the development of a community, see how risks are defined and monitored and how risk treatment plans become active.

Fig. 2. Metaphor-based visualization for the state of an online community.

Furthermore, ROBUST aims at providing alternative, metaphor-based visualization of online communities. One example is the fishtank visualization - a screenshot of an early prototype is shown in Figure 2. Using an animated submarine scenery the elements of the visualization depict various aspects of the state of health of an online community. The type and speed of fishes, the clarity of the water and the type and size of vegetation indicate role compositions, activity levels, and types of sentiments currently observed. Such visualizations will allow users to intuitively grasp the global state of an community with a single quick glance.

Management of online communities will become more and more important in the future, given the multiple forms of value generated by online communities. Pro-active support systems, such as the framework developed by the ROBUST project will make a substantial contribution to the performance of a community manager. Three further articles in this E-Letter will give deeper insights in the scalable social analytics for online communities, the approach for behaviour analysis and the risk management framework developed in the context of the ROBUST project.

The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 257859, ROBUST. Special thanks to Hugo Hromic and Toby Mostyn for providing the screenshots of the latest versions of the visualization tools.

[1] A. Che Alhadi, T. Gottron, J. Kunegis, and N. Naveed, “LiveTweet: Monitoring and Predicting Interesting Microblog Posts,” in ECIR’12: Procedings of the 34th European Conference on Information Retrieval, 2012, pp. 569–570.

[2] C. Boden, M. Karnstedt, M. Fernandez, and V. Markl, “Large-scale social-media analytics on stratosphere,” in WWW’13: Proceedings of the International Conference on World Wide Web, 2013.

[3] M. Rowe, M. Fernandez, S. Angeletou, and H. Alani, “Community analysis through semantic rules and role composition derivation,” Journal of Web Semantics, 2012. [Online]. Available:

[4] N. Naveed, T. Gottron, J. Kunegis, and A. Che Alhadi, “Bad News Travel Fast: A Content-based Analysis of Interestingness on Twitter,” in WebSci’11: Proceedings of the 3rd International Conference on Web Science, 2011.

[5] V. Belk, S. Lam, and C. Hayes, “Cross-community influence in discussion fora.” in Proceedings of the Sixth International Conference on Weblogs and Social Media, J. G. Breslin, N. B. Ellison, J. G. Shanahan, and Z. Tufekci, Eds. The AAAI Press, 2012.

[6] J. Preusse, J. Kunegis, M. Thimm, T. Gottron, and S. Staab, “Structural Dynamics of Knowledge Networks,” in ICWSM’13: Proceedings of the 7th International AAAI Conference on Weblogs and Social Media, 2013.

[7] F. Schwagereit, S. Sizov, and S. Staab, “Finding optimal policies for online communities with cosimo,” in Proceedings of the WebSci10: Extending the Frontiers of Society On-Line, April 26-27th, 2010, Raleigh, NC: US, 2010.

Dr Thomas Gottron
studied mathematics, computing science and business management at the Johannes Gutenberg University Mainz, Germany and the Glasgow University, UK. He received his PhD in computing science in 2008 at the University of Mainz. His research interests are in the area of Information Retrieval in the context of social networks and social media as well as semantic data, in the field of Web Science and in the analysis and management of online communities. His work is published in more than 50 top-level journals and conference proceedings. In 2011 he won the Billion Triple Track of the Semantic Web Challenge with SchemEX, a web-scale schema-level index for Linked Open Data. Since 2010 he is a PostDoc and research assistant at the Institute for Web Science and Technologies (WeST), University of Koblenz-Landau. Currently, he is the technical coordinator of the EC project ROBUST -- Risk and Opportunity management of huge-scale BUSiness communiTy cooperation.