We present our ongoing work on upgrading the Amsterdam Public Library's book database search capabilities. So far, users have had to input the exact book title and/or author name without any typos or misspellings in order to retrieve any results. This is in sharp contrast with the manner in which users typically use the interface: they frequently search for books on a particular topic, input the names of the characters, or even ask fully-fledged questions. The aim of this project is therefore to enable smart search in natural language based on book content. The initial focus is on the Dutch language, with the possibility of including English and other languages later. In the first phase of the project, we built a proof-of-concept knowledge graph from a sample of the existing tabular database and enriched the data with named entities extracted from book summaries. Based on this first step, a user query like "Heeft u boeken over de Tweede Wereldoorlog in Amsterdam?" would yield all books that mention both WW2 and Amsterdam. We are currently working on augmenting the knowledge graph with embeddings, which will enable us to retrieve semantically similar results. The final step of the research involves integrating our knowledge graph with a pre-trained large language model.
Estimating the remaining useful life (RUL) of an asset lies at the heart of prognostics and health management (PHM) of many operations-critical industries such as aviation. Mod- ern methods of RUL estimation adopt techniques from deep learning (DL). However, most of these contemporary tech- niques deliver only single-point estimates for the RUL without reporting on the confidence of the prediction. This practice usually provides overly confident predictions that can have severe consequences in operational disruptions or even safety. To address this issue, we propose a technique for uncertainty quantification (UQ) based on Bayesian deep learning (BDL). The hyperparameters of the framework are tuned using a novel bi-objective Bayesian optimization method with objectives the predictive performance and predictive uncertainty. The method also integrates the data pre-processing steps into the hyperparameter optimization (HPO) stage, models the RUL as a Weibull distribution, and returns the survival curves of the monitored assets to allow informed decision-making. We vali- date this method on the widely used C-MAPSS dataset against a single-objective HPO baseline that aggregates the two ob- jectives through the harmonic mean (HM). We demonstrate the existence of trade-offs between the predictive performance and the predictive uncertainty and observe that the bi-objective HPO returns a larger number of hyperparameter configurations compared to the single-objective baseline. Furthermore, we see that with the proposed approach, it is possible to configure models for RUL estimation that exhibit better or comparable performance to the single-objective baseline when validated on the test sets.
The research, supported by our partners, sets out to understand the drivers and barriers to sustainable logistics in port operations using a case study of drone package delivery at Rotterdam Port. Beyond the technical challenges of drone technology as an upcoming technology, it needs to be clarified how drones can operate within a port ecosystem and how they could contribute to sustainable logistics. KRVE (boatmen association), supported by other stakeholders of Rotterdam port, approached our school to conduct exploratory research. Rotterdam Port is the busiest port in Europe in terms of container volume. Thirty thousand vessels enter the port yearly, all needing various services, including deliveries. Around 120 packages/day are delivered to ships/offices onshore using small boats, cars, or trucks. Deliveries can take hours, although the distance to the receiver is close via the air. Around 80% of the packages are up to 20kg, with a maximum of 50kg. Typical content includes documents, spare parts, and samples for chemical analysis. Delivery of packages using drones has advantages compared with traditional transport methods: 1. It can save time, which is critical to port operators and ship owners trying to reduce mooring costs. 2. It can increase logistic efficiency by streamlining operations. 3. It can reduce carbon emissions by limiting the use of diesel engines, boats, cars, and trucks. 4. It can reduce potential accidents involving people in dangerous environments. The research will highlight whether drones can create value (economic, environmental, social) for logistics in port operations. The research output links to key national logistic agenda topics such as a circular economy with the development of innovative logistic ecosystems, energy transition with the reduction of carbon emissions, societal earning potential where new technology can stimulate the economy, digitalization, key enabling technology for lean operations, and opportunities for innovative business models.
The scientific publishing industry is rapidly transitioning towards information analytics. This shift is disproportionately benefiting large companies. These can afford to deploy digital technologies like knowledge graphs that can index their contents and create advanced search engines. Small and medium publishing enterprises, instead, often lack the resources to fully embrace such digital transformations. This divide is acutely felt in the arts, humanities and social sciences. Scholars from these disciplines are largely unable to benefit from modern scientific search engines, because their publishing ecosystem is made of many specialized businesses which cannot, individually, develop comparable services. We propose to start bridging this gap by democratizing access to knowledge graphs – the technology underpinning modern scientific search engines – for small and medium publishers in the arts, humanities and social sciences. Their contents, largely made of books, already contain rich, structured information – such as references and indexes – which can be automatically mined and interlinked. We plan to develop a framework for extracting structured information and create knowledge graphs from it. We will as much as possible consolidate existing proven technologies into a single codebase, instead of reinventing the wheel. Our consortium is a collaboration of researchers in scientific information mining, Odoma, an AI consulting company, and the publisher Brill, sharing its data and expertise. Brill will be able to immediately put to use the project results to improve its internal processes and services. Furthermore, our results will be published in open source with a commercial-friendly license, in order to foster the adoption and future development of the framework by other publishers. Ultimately, our proposal is an example of industry innovation where, instead of scaling-up, we scale wide by creating a common resource which many small players can then use and expand upon.
The drive to reduce the carbon intensity of the energy system has generated much interest in applying carbon-free fuels such as ammonia (NH3) in combustion systems. The high hydrogen density and well-established production processes make NH3 a valuable chemical energy carrier to address and sustain the energy shift toward renewable energy source integration. However, some difficulties can be highlighted in the NH3 practical application. The combustion of NH3 is prone to producing harmful nitric oxides. In addition, NH3 has lower reactivity than most hydrocarbon fuels, which makes ignition challenging. Also, admixing NH3 with highly reactive fuels such as DME will facilitate ignition. The partnerships of this proposal are very interested in applying renewable NH3 as fuel in combined heat and power engines, and this research proposal suggests simulating a dual-fuel engine with NH3 as its primary fuel. The results of this research will help determine the optimum operating conditions for performing an experimental study.
