Linköping Electronic Conference Proceedings
Not a member yet
1113 research outputs found
Sort by
The Influence of NegEx on ICD-10 Code Prediction in Swedish: How is the Performance of BERT and SVM Models Affected by Negations?
Get PDFClinical text contains many negated concepts since the physician excludes irrelevant symptoms when reasoning and concluding about the diagnosis. This study investigates the machine interpretation of negated symptoms and diagnoses using a rule-based negation detector and its influence on downstream text classification task. The study focuses on the effect of negated concepts and NegEx preprocessing on classifier performance for predicting ICD-10 gastro surgical codes assigned to discharge summaries. Based on the experiments, NegEx preprocessing resulted in a slight performance improvement for traditional machine learning model (SVM) and had no effect on the performance of the deep learning model KB/BERT
The concept of conditional method agreement trees with single measurements per subject
Get PDFThe concept of conditional method agreement is introduced and respective solutions are proposed to define homogeneous subgroups in terms of mean and variance of the differences in measurements
Time for using Machine Learning for Basal Insulin Dose Guidance for People with Type 2 Diabetes? Preliminary Results from a Systematic Review
Get PDFThe present systematic review aims to provide an overview and categorization of dose guidance methods that support basal insulin titration for people with type 2 diabetes. At the time of writing, quality assessment of the included articles is ongoing
Curation Criteria for Multimodal and Multilingual Data: a Mixed Study within the QUEST Project
Get PDFA user survey was conducted and expert interviews within the ongoing QUEST project to get an impression of the needs of users and researchers who are working with multimodal and multilingual linguistic corpora. This contribution describes the design and results of the mixed study, whose main goal is to improve the reuse potential of these resources, and to identify concrete topics which are important for the curation of such data
New Ciphers and Cryptanalysis Components in CrypTool 2
Get PDFIn this paper, we discuss new additions (components) for cryptography and cryptanalysis added to the CrypTool 2 (CT2) software over the course of the last two years. We mainly focus on components for classical and historical ciphers, but also keep an eye on other updates of CT2, e.g. the CrypConsole, which allows the users to execute CT2 workspaces in the Windows command prompt. The Chaocipher as well as Josse’s cipher were added to CT2. The Symbol Cipher component already allows the user to create musical ciphers. We implemented a new Playfair Analyzer as well as a Josse Cipher Analyzer. The Enigma Analyzer component has been rewritten completely and now features six different computerized attacks on Enigma. Two historical ciphers were successfully deciphered with the help of CT2: five ciphertexts from and to the Holy Roman Emperor Maximilian II as well as the Ramanacoil transcript. Both ciphers from the 16th and 17th century were analyzed and deciphered using the Homophonic Substitution Analyzer component as well as the substitution component. Finally, we take a brief look at how CT2 is used for teaching and e-learning cryptology
Deciphering a Short Papal Cipher from 1721
Get PDFAs part of the DECRYPT project, hundreds of enciphered papal letters from the 16th, 17th, and 18th centuries were deciphered, and dozens of keys were recovered. Several ciphertexts remained unsolved, despite the use of sophisticated computerized algorithms, and were offered as public challenges in 2019. One of those is a short letter sent by the nuncio in Bruxelles to Rome, on October 9, 1721. It consists of groups of digit codes, separated by commas. After improving the algorithm, and with some manual work, the authors were able to recover most of the key and of the plaintext. In this article, they present the method they used to recover the key, and the decrypted message
The TRANSCRIPT Tool for Historical Ciphers by the DECRYPT Project
Get PDFTRANSCRIPT is a web-based tool1 for creating transcriptions for scanned images of historical manuscripts. The tool is interactive and leverages pre-trained image processing algorithms, therefore implementing a human-in-the-loop artificial intelligence. Different algorithms may be used in every step of the process: line and symbol segmentation, clustering and symbol recognition. However, at each step the user can manually intervene, clean up or enrich the results of algorithmic image processing. We present here the current work-in-progress version of the TRANSCRIPT tool
Study on BEV concept design based on data driven approach
Get PDFThis paper researched the Battery EV concept design based on the data driven model. To determine the performance of BEV in the concept stage, a database was established through market research, and a data driven model was created to derive the target performance and specifications based on the database. To verify the results of the data driven model, the BEV model was generated, and the derived specifications were set. After that, the target performance was confirmed through simulation and detailed specifications were derived
Modelling and Optimal Design of Gas Engine CCHP System in Hospital
Get PDFThe combined heating and power (CHP) system and the combined cooling, heating and power (CCHP) system have attracted great attention during the last decade. However, many CHP systems don’t perform well in the actual operation. This paper presents a complete hierarchical modeling tool of the gas engine CHP/CCHP system which is built on the software Dymola. Meanwhile, a gas-engine CHP hybrid energy system serving a hospital in Shanghai is studied as a case. To validate the accuracy of newly-built models, the operating data of the CHP part of the system in 2017 is compared with the simulation results, it is found that the minimum error is 2.1%, and the maximum error is 7.0%. Then, the original gas engine CHP hybrid energy system is reconstructed to a gas engine CCHP system. To analyze the feasibility of the optimal design, the conventional energy supply system which was used in the hospital before 2013, the original gas engine CHP hybrid energy system and the optimized gas engine CCHP system are modeled and simulated. From the simulation results, it is found that the primary energy ratio is increased from 72.55% to 133.37%, the payback period of investment is decreased from nearly 11.8 years to 3.9 years, and the CO2 emissions reduction rate is increased from 4.83% to 93.72%. Therefore, the optimization scheme is feasible
The Teacher-Student Chatroom Corpus version 2: more lessons, new annotation, automatic detection of sequence shifts
Get PDFThe first version of the Teacher-Student Chatroom Corpus (TSCC) was released in 2020 and contained 102 chatroom dialogues between 2 teachers and 8 learners of English, amounting to 13.5K conversational turns and 133K word tokens. In this second version of the corpus, we release an additional 158 chatroom dialogues, amounting to an extra 27.9K conversational turns and 230K word tokens. In total there are now 260 chatroom lessons, 41.4K conversational turns and 363K word tokens, involving 2 teachers and 13 students with seven different first languages. The content of the lessons was, as before, guided by the teacher, and the proficiency level of the learners is judged to range from B1 to C2 on the CEFR scale. Annotation of the dialogue continued with conversational analysis of sequence types, pedagogical focus, and correction of grammatical errors. In addition, we have annotated fifty of the dialogues using the Self-Evaluation of Teacher Talk framework which is intended for self-reflection on interactional aspects of language teaching. Finally, we conducted machine learning experiments to automatically detect shifts in discourse sequences from turn to turn, using modern transfer learning methods with large pretrained language models. The TSCC v2 is freely available for research use