The Internet of Things (IoT) aims to interconnect billions of heterogeneous devices across diverse applications. To ensure interoperability, OM2M stands out as a popular open-source middleware platform for integrating heterogeneous IoT devices; however, its deployment complexity and limited documentation pose significant barriers to broader adoption. Existing implementations of OM2M often rely on centralized architectures, which can lead to a single point of failure, compromising the IoT system's resilience and scalability. In this paper, we address this challenge by proposing a lightweight, automated approach to install and configure OM2M that simplifies both the deployment process and end-node interactions. Our solution allows for a decentralized implementation of OM2M, mitigating the risks associated with centralized configurations. The proposed solution uses Docker-based containerization compatible across IoT devices with various central processing unit (CPU) architectures. Additionally, we provide a Python-based abstraction, implemented in both CPython and MicroPython, which streamlines the creation, retrieval, and management of oneM2M resources on an OM2M common service entity (CSE). Implementation results show that our containerized approach reduces CPU usage, power consumption, and memory usage compared to the bash-script-based method. These improvements highlight its efficiency in resource-constrained IoT scenarios. The proposed solution is available open-source on GitHub to encourage broader adoption, collaboration, and long-term viability within the IoT community.

International audience

The Web of Data has grown explosively over the past few years, and as with any dataset, there are bound to be invalid statements in the data, as well as gaps. Natural Language Processing (NLP) is gaining interest to fill gaps in data by transforming (unstructured) text into structured data. However, there is currently a fundamental mismatch in approaches between Linked Data and NLP as the latter is often based on statistical methods, and the former on explicitly modelling knowledge. However, these fields can strengthen each other by joining forces. In this position paper, we argue that using linked data to validate the output of an NLP system, and using textual data to validate Linked Open Data (LOD) cloud statements is a promising research avenue. We illustrate our proposal with a proof of concept on a corpus of historical travel stories.

We present QAnswer, a Question Answering system which queries at the same time 3 core datasets of the Semantic Web, that are relevant for end-users. These datasets are Wikidata with Lexemes, LinkedGeodata and Musicbrainz. Additionally, it is possible to query these datasets in English, German, French, Italian, Spanish, Pourtuguese, Arabic and Chinese. Moreover, QAnswer includes a fallback option to the search engine Qwant when the answer to a question cannot be found in the datasets mentioned above. These features make QAnswer as the first prototype of a Question Answering System over a considerable part of the LOD cloud.