Anne Liret - Academia.edu (original) (raw)

Papers by Anne Liret

Lecture notes in computer science, 2024

Zenodo (CERN European Organization for Nuclear Research), Sep 30, 2023

HAL (Le Centre pour la Communication Scientifique Directe), Jun 6, 2016

Zenodo (CERN European Organization for Nuclear Research), Dec 22, 2022

Digital supply chain transformation: emerging technologies for sustainable growth

This chapter considers how Artificial Intelligence (AI) can be used to optimise assets and invent... more This chapter considers how Artificial Intelligence (AI) can be used to optimise assets and inventory. It looks at how a business such as BT deals with the key resources that service organizations such as telecommunications companies maintain: their assets and inventories. In the discussion the authors address both strategic and operational dimensions of the deployment challenge for such companies. From a strategic perspective there is a need to deploy fixed assets for optimal performance; at the same time, from an operational perspective, there is a need to replenish inventory to be able to deliver services in line with customer service level agreements. This creates a ‘combinatorial optimization problem’ which, it is suggested, makes AI a useful technology for solving such problems for operational use.

arXiv (Cornell University), Apr 4, 2024

Retrieval-Augmented Generation (RAG) enhances Large Language Model (LLM) output by providing prio... more Retrieval-Augmented Generation (RAG) enhances Large Language Model (LLM) output by providing prior knowledge as context to input. This is beneficial for knowledge-intensive and expert reliant tasks, including legal question-answering, which require evidence to validate generated text outputs. We highlight that Case-Based Reasoning (CBR) presents key opportunities to structure retrieval as part of the RAG process in an LLM. We introduce CBR-RAG, where CBR cycle's initial retrieval stage, its indexing vocabulary, and similarity knowledge containers are used to enhance LLM queries with contextually relevant cases. This integration augments the original LLM query, providing a richer prompt. We present an evaluation of CBR-RAG, and examine different representations (i.e. general and domain-specific embeddings) and methods of comparison (i.e. inter, intra and hybrid similarity) on the task of legal question-answering. Our results indicate that the context provided by CBR's case reuse enforces similarity between relevant components of the questions and the evidence base leading to significant improvements in the quality of generated answers.

CERN European Organization for Nuclear Research - Zenodo, Mar 22, 2022

CERN European Organization for Nuclear Research - Zenodo, Dec 3, 2021

D1.1. Ontology requirements completed [RGU] D2.1a. So... more D1.1. Ontology requirements completed [RGU] D2.1a. Software development plan [UCM] D5.1a. Systematic review of evaluation measurements [RGU/BT] D6.1. iSee web and community network ready. [BT] D7.1a. Project Handbook [UCM] D7.2. Quality Plan [BTF] D7.4b. Consortium agreement (UCM)

2017 Computing Conference, 2017

2016 IEEE 28th International Conference on Tools with Artificial Intelligence (ICTAI), 2016

This paper proposes a constraint optimization model accepting different network topologies, site ... more This paper proposes a constraint optimization model accepting different network topologies, site functions and distribution/transfer policies, applies it to a real case study or closed-loop supply chains in telecommunications services, and compares two approaches-a mixed-integer programming and a metaheuristics-to solve this problem. Its cost function is a linear combination of storage, transport, backorder and repair costs. We report experiments on pseudo random instances designed to evaluate plan quality and impact of cost weightings. Experiments validate approaches and compare our 2 methods. Finally, we discuss the possible extensions of the model to fit other specific cases and create interest among supply chain experts.

9th IFAC Conference Manufacturing Modelling, Management and Control MIM 2019, Aug 28, 2019

We present a metaheuristic for planning the distribution of items in closed-loop supply chains. T... more We present a metaheuristic for planning the distribution of items in closed-loop supply chains. This metaheuristic composes sequences of transfer and repair actions to generate plans iteratively. It uses a local search algorithm based on an efficient data structure to construct and select improving sequences at each step. An experimental comparison with a mixed integer programming approach shows its scalability and robustness on a variety of instances. We also study and discuss the ability to support different distribution policies.

Transportation Research Part E: Logistics and Transportation Review, 2022

Lecture Notes in Computer Science, 2018

Within any sufficiently expertise-reliant and work-driven domain there is a requirement to unders... more Within any sufficiently expertise-reliant and work-driven domain there is a requirement to understand the similarities between specific work tasks. Though mechanisms to develop similarity models for these areas do exist, in practice they have been criticised within various domains by experts who feel that the output is not indicative of their viewpoint. In field service provision for telecommunication organisations, it can be particularly challenging to understand task similarity from the perspective of an expert engineer. With that in mind, this paper demonstrates a similarity model developed from text recorded by engineer's themselves to develop a metric directly indicative of expert opinion. We evaluate several methods of learning text representations on a classification task developed from engineers' notes. Furthermore, we introduce a means to make use of the complex and multi-faceted aspect of the notes to recommend additional information to support engineers in the field.

Proceedings of SW20 The OR Society Simulation Workshop, 2020

Tuesday, March 23 Day 3 Wednesday, March 24 Day 4 Thursday, March 25 Day 5 Friday, March 26 Netwo... more Tuesday, March 23 Day 3 Wednesday, March 24 Day 4 Thursday, March 25 Day 5 Friday, March 26 Networking break: Speed networking Lunch Break Lunch (1st 30 mins + networking (2nd half hour):(Trying to) Find calm in the chaos (Informal discussion in groups) Lunch (1st 30 mins + networking (2nd half hour): Lunchtime breakout rooms to discuss a simulation topic of interest Lunch (1st 30 mins + networking (2nd half hour)-3 Lunchtime breakout discussion rooms: 1) Your career in Simulation+ 2) WORAN + 3)ORS Orientation

Artificial Intelligence XXXIV, 2017

A decision support system is designed in this paper for supporting the adoption of green logistic... more A decision support system is designed in this paper for supporting the adoption of green logistics within scheduling problems, and applied to real-life services cases. In comparison to other green logistics models, this system deploys time-varying travel speeds instead of a constant speed, which is important for calculating the CO2 emission accurately. This system adopts widely used instantaneous emission models in literature which can predict second-by-second emissions. The factors influencing emissions in these models are vehicle types, vehicle load and traffic conditions. As vehicle types play an important role in computing the amount of emissions, engineers vehicles number plates are mapped to specified emission formulas. This feature currently is not offered by any commercial software. To visualise the emissions of a planned route, a Heat Map view is proposed. Furthermore, the differences between minimising CO2 emission compared to minimising travel time are discussed under different scenarios. The field scheduling problem is formulated as a vehicle routing and scheduling problem, which considers CO2 emissions in the objective function, heterogeneous fleet, time window constraints and skill matching constraints, different from the traditional time-dependent VSRP formulation. In the scheduler, this problem is solved by metaheuristic methods. Three different metaheuristics are compared. They are Tabu search algorithms with random neighbourhood generators and two variants of Variable Neighbourhood search algorithms: variable neighbourhood descent (VND) and reduced variable neighbourhood search (RVNS). Results suggest that RVNS is a good trade-off between solution qualities and computational time for industrial application.

Within the service providing industries, field engineers can struggle to access tasks which are s... more Within the service providing industries, field engineers can struggle to access tasks which are suited to their individual skills and experience. There is potential for a recommender system to improve access to information while being on site. However the smooth adoption of such a system is superseded by a challenge for exposing the human understandable proof of the machine reasoning.With that in mind, this paper introduces an explainable recommender system to facilitate transparent retrieval of task information for field engineers in the context of service delivery. The presented software adheres to the five goals of an explainable intelligent system and incorporates elements of both Case-Based Reasoning and heuristic techniques to develop a recommendation ranking of tasks. In addition we evaluate methods of building justifiable representations for similarity-based return on a classification task developed from engineers' notes. Our conclusion highlights the trade-off between p...