Personal data ownership and interoperability for decentralized social Web applications are currently two debated topics, especially when taking into consideration the aspects of privacy and access control. To increase data ownership, users should have the freedom to choose where their data resides and who is allowed access to it by decoupling data storage from the application that consumes it. Through CIMBA, we propose a decentralized architecture based on Web standards, which puts users back in control of their own data.
Recently, we have witnessed a dramatic increase in the number of social Web applications. These applications come in di erent forms and o er di erent services such as social networks, content management systems (CMS), bug trackers, blogging tools, or collaboration services in general.
A common practice, speci c to most Web services, is to centralize user resources thus becoming so-called data silos. Often when adhering to online services people usually end up creating dedicated local accounts, which ties and limits users to particular services and/or resources. A solution to data silos can be achieved through decentralization, where users are free to host their data wherever they want, and then use several Web apps to consume and manage the data. In the following section we will discuss how our decentralized architecture plays an important role in achieving true data ownership for users. structured according to each speci c application and it only has meaning within the context of those applications. Moreover, this practice also forces a tight coupling between backends and applications consuming the data (cf. Fig.1 a).
The proposed architecture uses the Resource Description Framework (RDF) [
Another advantage is that users are no longer locked into a silo because
of their social connections (cf. Fig.2 b). Web apps reuse the user's social graph,
which is also located on the data manager. The data manager is a generic Linked
Data personal data server, which implements the Linked Data Platform
speci cation [
Our architecture uses WebID [
CIMBA - Client-Integrated MicroBlogging Architecture The name CIMBA stands for Client-Integrated MicroBlogging Application. It provides users with the power of having their own blog combined with the ease of using Twitter. CIMBA was written in Javascript, using the AngularJS4 framework. The source code is publicly availabled on GitHub5, and a running online demo can be accessed by visiting http://cimba.co.
Compared to Twitter, CIMBA users are not stuck with a single feed or timeline, but instead they can create multiple Channels and use them as categories for their posts (i.e. main, work, family, etc.). Access control can be set per channel as well as per post, though policies for posts will override those set per channel (i.e. having a private post in a public channel).
Figure 3 displays an overview of the architecture behind CIMBA. Users Alice and Bob each have their own personal data managers, which hold the posts data, con guration les as well as their personal WebID pro les.
Accessing CIMBA simply means loading all the necessary HTML, Javascript and CSS les from the application server into the user's browser. From that moment on, the application which now runs in the browser will communicate directly with the user's personal data manager. The location of the personal data manager is found by "faking" a WebID-TLS authentication process, with the purpose of nding the user's WebID and implicitly, the WebID pro le. There is no actual need to authenticate the user to CIMBA, since all requests for data are authenticated by the user's personal data store.
Once the WebID pro le is found, the app follows a series of links to discover useful information about the user, such as a generic Linked Data server from where the app can store and retrieve resources. CIMBA stores all the application data on the user's personal data manager, in a workspace dedicated to microblogging. Microblogging data are stored using the SIOC ontology. 4 https://angularjs.org/ 5 https://github.com/linkeddata/cimba
To read what other people write, users can subscribe to their channels. The list of subscriptions is also expressed using the SIOC vocabulary and it is stored on the user's server. 4
Conclusions and future work Our proposed decentralized architecture o ers signi cant bene ts compared to current Web apps, both in terms of data ownership, privacy, as well as interoperability. Being fully decoupled from the backend, Web apps can be easily forked and improved by anyone with access to the source code, thus spurring innovation and creativity. At its current stage, CIMBA su ers from scalability issues, though we are very close to overcoming these issues.