Since the Web 2.0 era, the Internet is a huge content editing place on which users collaborate. Thousand of people can edit this shared document. However, current consistency maintenance algorithms are not adapted to massive collaborative updating involving large number of contributors and a high velocity of changes. This year we studied collaborative editing user behaviour and started to work on an optimised solution for sequence CRDTs. Version control systems such as Git became very widespread in the open-source community. In these collaborative systems, conflict resolution that arise during synchronisation of parallel changes might become a burden for the user. We analysed concurrency and conflicts in Git repository of four projects: Rails, IkiWiki, Samba and Linux Kernel. We analysed the collaboration process of these projects at specific periods revealing how change integration and conflict rates vary during the project development life-cycle. Our study suggests that developers should use more intensively awareness mechanisms close to release dates where changes integration rate is higher. We also discussed the mechanism adopted by Git to consider concurrent changes made on two adjacent lines as conflicting. Based on the high rate of false positives of this mechanism, our study suggests that Git should reconsider signalling adjacent line conflicts inside the source code files . Sequence Conflict-free Replicated Data Types (CRDTs) allow one to replicate and edit, without any kind of coordination, sequences in distributed systems. To ensure convergence, existing works from the literature add metadata to each element but they do not bind its footprint, which impedes their adoption. Several approaches were proposed to address this issue but they do not fit a fully distributed setting. We started to work on the design and validation of a fully distributed renaming mechanism, setting a bound to the metadata’s footprint . Addressing this issue opens new perspectives of adoption of these CRDTs in distributed applications.

Trust between users is an important factor for the success of a collaboration. Users might want to collaborate only with those users they trust. We are interested in assessing users trust according to their behaviour during collaboration in a large scale environment. We studied the trust assessment problem and designed a computational trust model for collaborative systems . We also studied how to predict the trust relation between users that did not interact in the past. Given a network in which the links represent the trust/distrust relations between users, we aimed to predict future relations. We proposed a link-sign prediction algorithm  that does not require full graph information, is suitable for dynamic networks and takes into account the creation time of the links in the network. Our solution combines state-of-the-art techniques in natural language processing (Doc2Vec ) and deep learning (Recurrent Neural Networks with Long-Short Term Memory ) with the random walk graph sampling . Our algorithm outperforms state-of-the-art approaches on real world signed directed social network datasets. In distributed collaborative systems, participants maintain a replicated copy of shared documents. They edit their own copy and then share their modifications without any coordination. Copies follow successions of divergence and convergence. Convergence is a liveness property of collaborative systems. Some malicious participants may find an advantage to make the collaboration fail. To that end, they can preclude convergence of the copies. To protect convergence of copies, participants can exploit an authenticated log of modifications. New participants have to retrieve the entire log in order to contribute. Unfortunately, the cost of joining a collaboration increases with the size of this log. Causal Stability allows to prune authenticated logs in a static collaborative group without any malicious participants. We tailored Causal Stability to dynamic groups in the presence of malicious participants. We also proposed a mechanism to verify the consistency of a pruned log and a mechanism to authenticate a snapshot from a pruned log . Public key server is a simple yet effective way of key management in secure end-to-end communication. To ensure the trustworthiness of a public key server, CONIKS employs a tamper-evident data structure on the server and a gossiping protocol among clients in order to detect compromised servers. However, due to lack of incentive and vulnerability to malicious clients, a gossiping protocol is hard to implement in practice. Meanwhile, alternative solutions such as EthIKS are too costly. We proposed Trusternity , , an auditing scheme relying on Ethereum blockchain that is easy to implement, inexpensive to operate and resilient to malicious clients. We also conducted an empirical study of system behaviour in face of attacks and proposed a lightweight anomaly detection algorithm to protect clients against such attacks.

Sharing information between responders is important during crisis management response. Tools and platforms are eagerly developed for that purpose. They are supposed to support people and help them to build a shared situation awareness. However as the scale of crisis increases and as more and more organisations are involved, people get reluctant to use them to share their data. They prefer to rely on one to one communication tools like phones or text. This is why we are studying how these collaborative platforms impact the work of responders positively or negatively. We want to know why most of the time they don't want to use them for their original purpose. We studied reports on past incidents , and conducted extensive analysis of the use of existing systems (e.g. the French platform CRISORSEC) through interviews, observation and data analysis. 

Cloud computing providers do not help consumers to use optimally the available resources. Several approaches have been proposed that take benefit from the elasticity of the Cloud, starting and stopping virtual machines on demand. They suffer from several shortcomings. Often they consider only one objective, the reduction of the cost, or a level of quality of service. We proposed to optimise two conflicting objectives, the number of migrations of tenants that is helpful to reach the optimal cost and the cost incurred considering a set of resources. Our approach allows to take into account the multi-tenancy property and the Cloud computing elasticity, and is efficient as shown by an extensive experimentation based on real data from Bonita BPM customers. In the continuation of our previous work we proposed and validated a more efficient algorithm for elastic execution of processes in the cloud . To ensure a realistic validation, we collaborated with colleagues from the University of Lugano to set up a benchmarking platform in order to evaluate the impact of migration in a multi-tenant setting. This allowed us to execute reproducible experiments and to validate our hypothesis regarding the effect of migration and the parameters that affect them . This platform is now an asset that can be used for all kinds of live migration experiments of software architectures.

The lack of trust in cloud organisations is often seen as braking forces to SaaS developments. This work proposes an approach which supports a trust model and a business process model in order to allow the orchestration of trusted business process components in the cloud. The contribution is threefold and consists in a method, a model and a framework. The method categorises techniques to transform an existing business process into a risk-aware process model that takes into account security risks related to cloud environments. These techniques are partially described in the form of constraints to automatically support process transformation. The model formalises the relations and the responsibilities between the different actors of the cloud. This allows to identify the different information required to assess and quantify security risks in cloud environments. The framework is a comprehensive approach that decomposes a business process into fragments that can automatically be deployed on multiple clouds. The framework also integrates a selection algorithm that combines the security information of cloud offers and of the process with other quality of service criteria to generate an optimised configuration. It is implemented in a tool to assess cloud providers and decompose processes. Rooted in past years work, we are contributing this year at the methodological and framework levels in two directions:

Business Process Management (BPM) is concerned with continuously enhancing business processes by adapting a systematic approach that enables companies to increase the performance of their existing business processes and achieve their business goals. Business processes are generally considered as blind, stateless and reactive. This means that in each business process execution we do not take into consideration either the results from last process instances nor the context (for most cases). The rise of new technologies such as big and fast data, cloud computing, Internet of Things (IoT), etc, implies new business process scheduling problems. They are linked to limited resources (human and/or machine) or the need to use resources in an optimal and exible way. In order to avoid either under-provisioning (when there is an underestimation for the needed resources, business processes may not be executed) or over-provisioning (the resources planned in advance to cover peak times demands were not used in non-peak time) and also to take into consideration the priority level of each business process instances.