The CAASC project considers a supply chain where independent partners accept to share in a cloud platform a series of information about their mid-term plans. Maintenance and misalignment of these plans is a major source of problems (stock-outs, over-stocks, distribution over-costs, irrelevant capacities) in supply chains. Shared information concern agreed plans, but also the current status and current forecasts associated to sales, production, inventory, transports. In particular, CAASC is interested in the detection of deviations between the agreed partners’ plans and the current situation in order to look for, if required, supply chain agile adaptations to these deviations. CAASC takes advantage of the software and test infrastructures developed for the H2020 C2NET project. But, while C2NET was interested in the import of data, their integration into a supply chain model and the detection of deviations, CAASC is focusing on the analysis of deviations and the development of agility capabilities. One major add-on of CAASC is the management of uncertainties in the monitoring process of a plan in order to reduce the nervousness of decisions, counter the bullwhip effect or avoid an irrelevant protection of decoupling points. As a consequence, CAASC develops three research axis: The first axis concerns the identification and quantification of uncertainties. The point is to take advantage of the flow of data in the solution, and the capabilities of automated learning algorithms. The goal is to model effective uncertainties in a supply chain: typify, classify, quantify or discover emergent uncertainties, while selecting the appropriate modelling approach (probabilities, possibilities, uncertain probabilities, ). This axis takes advantage of the knowhow of Linagora for machine learning algorithms, the experience in decision under uncertainty of IRIT/ADRIA and the knowledge in supply chain risks management of the whole consortium. The second axis studies the projection of a series of uncertainties and deviations on a plan. It evaluates the capability of an actor to control the effects of uncertainties and deviations on his perimeter. It results in the identification of risks for the other partners. While bibliography is systematically focusing on one type of uncertainty, the goal here is to integrate various representations of uncertainties, in order to result in a more realistic risks characterisation. Third, for critical risks anticipating adaptation strategies are mandatory. More than re-planning algorithms, we propose to provide a compact representation of risks management strategies for supply chain planning. This totally new approach in the domain of supply chain planning, based on the principles of knowledge compilation, will guaranty finding adaptations in polynomial time when risks occur. CAASC considers two types of case studies adapted from C2NET and integrating supply chain uncertainties and risks. The first is a series of scenarios of various sizes that result from real data extracted from the Pierre Fabre Group information systems. They enable measuring the scalability of the solution. The second type is a serious game that enables to assess the dynamic behaviours. On these use cases, various managerial effects are looked for: the formalisation of the decision makers behaviours, the impacts of the proposed tools on the capacity of people to collaborate, and more globally, the organisational changes. Finally, the tools and services developed during this project aim at being integrated in the OpenPaaS solution, the cloud collaborative open-source platform provided by Linagora. The objective is to develop differentiating services for industrial companies. Overall, the funding will support 2 PhDs, a high scientific integration with Linagora and an industrial validation by the Pierre Fabre company.

It is becoming increasingly realistic to exploit transcriptions of spoken data for tasks that require comprehension of what is said in a conversation. SUMM-RE will combine expertise in theories of discourse interpretation with recent developments in distant supervision to improve the automatic production of meeting summaries and minutes from spoken data. State of the art approaches to abstractive summarization treat discourse as a mere linear sequence of utterances. SUMM-RE posits that by exploiting information about discourse relations and the rich structures determined by relations between utterances, we can significantly improve models for abstractive summarization. A major hurdle to developing more sophisticated models of discourse structure for spoken, multiparty conversation is a lack of appropriate training data. SUMM-RE will address this problem in two ways. First, it will create a new and unique corpus of meeting-like interactions in French. Second, it will label this corpus and a large corpus of meeting-like interactions in English for discourse structure. The annotation approach will extend recent developments in distant supervision to develop labelling functions that can be used to automatically label large amounts of data. This approach has the very attractive advantage of harnessing linguistic expertise while keeping manual annotation to a minimum. The automatically annotated data will be used to improve algorithms for both short topic summaries and more detailed meeting minutes. These algorithms in turn will be integrated into the lead partner's (LINAGORA's) semi-automatic summarization tool to significantly improve the output for its users. All project results (corpus and algorithms) will be released under an open-source license as a part of LINAGORA's LinTo/Conversation Manager offer.