• 2020-2024close
• Open Access close
• Closed Access close
• Open Source close
• Research data close
• Other research products close
• Other ORP type close
• Digital Humanities and Cultural Her... close

El presente trabajo evalúa la producción de la traducción automática del asistente virtual de Apple, Siri. También da un breve repaso al estado de la cuestión, que explora como Siri 'aprende' nuevos idiomas gracias al procesamiento del lenguaje natural y utiliza esta información para su traductor automático. Seguidamente, se prueba el nivel de rendimiento del traductor de Siri en la combinación lingüística inglés (Reino Unido) > español a través del dictado de textos mediante el procesamiento de voz, divididos por ámbitos temáticos, niveles de especialización y códigos lingüísticos. La finalidad es comprender si el rendimiento del sistema es adecuado para cubrir las necesidades del usuario. En el apartado práctico se determina un esquema de clasificación de errores basado en aproximaciones elaboradas por expertos, y se presenta un informe detallado con los errores más recurrentes y los puntos con una mayor relevancia. El present treball avalua la producció de la traducció automàtica de l'assistent virtual d'Apple, Siri. També fa un breu repàs de l'estat de la qüestió, que explora com Siri 'aprèn' nous idiomes gràcies al processament del llenguatge natural i utilitza aquesta informació per al seu traductor automàtic. Seguidament, es prova el nivell de rendiment del traductor de Siri en la combinació lingüística anglès (Regne Unit) > espanyol dictant textos mitjançant el comandament de veu, dividits per àmbits temàtics, nivells d'especialització i codis lingüístics. La finalitat és comprendre si el rendiment del sistema és prou adequat per cobrir les necessitats de l'usuari. A l'apartat pràctic es determina un esquema de classificació d'errors basat en aproximacions elaborades per experts, i es presenta un informe detallat amb tots els errors més recurrents i els punts amb una major rellevància. This paper evaluates the output of the automatic translation of Apple's virtual assistant, Siri. It also makes a brief review of the state of the art, understanding how Siri 'learns' new languages by the processing of natural language and uses this information for machine translation. Then, it tests the performance level of Siri's translator in the English (UK) > Spanish language combination by dictating texts by voice command, divided by subject areas, levels of specialization and language codes. The purpose is to understand if the system's performance is adequate enough to meet the user's needs. Furthermore, it is determined a scheme of classification of errors based on previous approaches, and it is presented a detailed report with all the most frequent mistakes and other relevant aspects.

This thesis explores the potential of Large Language Models (LLMs) to assist in the technical documentation process at Saab. With the increasing complexity and regulatory demands on such documentation, the objective is to investigate advanced natural language processing techniques as a means of streamlining the creation of technical documentation. Although many standards exist, this thesis particularly focuses on the standard ASD-STE100, Simplified Technical English abbrv. STE, a controlled language for technical documentation. STE's primary aim is to ensure that technical documents are understandable to individuals regardless of their native language or English proficiency. The study focuses on the implementation of Direct Preference Optimization (DPO) and Supervised Instruction Fine-Tuning (SIFT) to refine the capabilities of LLMs in producing clear and concise outputs that comply with STE. Through a series of experiments, we investigate the effectiveness of LLMs in interpreting and simplifying technical language, with a particular emphasis on adherence to STE standards. The study utilizes a dataset comprised of target data paired with synthetic source data generated by a LLM. We apply various model training strategies, including zero-shot performance, supervised instruction fine-tuning, and direct preference optimization. We evaluate the various models' output using established quantitative metrics for text simplification and substitute human evaluators with company internal software for evaluating adherence to company standards and STE. Our findings suggest that while LLMs can significantly contribute to the technical writing process, the choice of training methods and the quality of data play crucial roles in the model's performance. This study shows how LLMs can improve productivity and reduce manual work. It also looks at the problems and suggests ways to make technical documentation automation better in the future.

Sheep predominate the Early Neolithic faunal assemblages in the Iberian Peninsula. Their exploitation for meat and milk production made them key to the economy of these early farming societies. Management of sheep breeding season and feeding in the context of the local environment were decisive in obtaining these livestock products. This work focuses on these aspects through stable isotope and dental microwear analyses on sheep teeth from the cave of Chaves (Huesca, Spain). The results show the existence of "out of season" (autumn/early winter) sheep births in the Early Neolithic, contrasting significantly with spring lambing prevailing in Neolithic husbandries elsewhere in Europe and confirming the antiquity of a western Mediterranean characteristic in this regard. Furthermore, little changes in sheep diet throughout the year have been documented, as far as could be evidenced from stable carbon isotope ratios and dental microwear. Only two individuals showed higher variability in diet on a seasonal scale with possible contribution of C-4 plants, possibly from grazing in the valley steppes at lower altitudes. Overall the results suggest good adaptation of sheep to the Pyrenean mid-altitude environment and strong zootechnical knowledge of the earliest shepherds in this area.

Efforts to “greenshift” (Backe) digital games have been hindered, in part, by a gap between game development practices and critical game studies. To date, plant modeling for games has largely focused on imitating visual patterns rather than implementing dynamic ecological processes, which has shaped and ultimately limited the arguments that games are able to make about the natural world. Propagate, an individual-based plant growth plugin for Unreal Engine, draws from both ecological modeling and game criticism to deepen the relationship between play and ecology. The project offers a speculative approach to cultivating persistent digital environments that vary over time, are delicately interconnected, and exist in dialogue with players. Approaching game worlds as living, animate environments instead of painstakingly crafted backdrops is a step towards redistributing narrative focus and agency from players to the spaces they inhabit.

In 1920s Sweden the concept of Folkhemmet was born and described as a concept and an idea of the modern Swedish well-fare state where the nation was considered a family. Through governmental founding, a new modern Sweden was planned. The first suburbs was introduced in 1930 and the 1950s the next project was finished, the so-called ABC-staden (ABC-town in English). One of the suburbs belonging to the ABC-town project was Rågsved, a place which is associated with the Swedish punk movement and the cultural association of Oasen. In 1965 the next building project was introduced, now called Miljonprogrammet. Miljonprogrammet became criticized for being a constant working place and isolating citizens in the suburbs, leading to segregation. A furthermore consequence of the segregation was when working-class families got a better economy and left the traditional apartment housing for newly produced town halls it opened for low-income families to move in. Meanwhile, the media framing of the suburb also began to change. The narrative of the suburb had previously been a hope for a bright better future for families to live in, but it later changed to be a place for criminality, under-employment, alcoholism, and substance abuse. Teenagers and young adults stuck at the bottom of the economic chain were affected worst by this change. This also occurred with the oil crisis in the late 1970s, making the end of the Swedish record years. The Swedish punk movement was introduced in 1976 inspired by the British and American punk scene. The movement established itself in the suburbs among young adults who did not feel like they were heard. The movement was, in difference to previous youth cultures, political which affected the message of the movement. In the capital of Stockholm, punk rockers found themselves wanting something more, a place outside the centre of the city where they all could have a place to be. This project later resulted in the cultural association of Oasen in Rågsved. In 1979 after the city council of Vantör decided to throw out the association they answered by occupying the locale. They demanded a place to feel at home in the suburb they lived in. This thesis is discussing the modern Swedish suburb using the Swedish punk movement as a historic example. The thesis’s theoretical framework focuses on intersectionality, place, space, and journalistic framing. 

Data ----- We have collected English-Odia parallel data for the purposes of NLP research of the Odia language. The data for the parallel corpus was extracted from existing parallel corpora such as OdiEnCorp 1.0 and PMIndia, and books which contain both English and Odia text such as grammar and bilingual literature books. We also included parallel text from multiple public websites such as Odia Wikipedia, Odia digital library, and Odisha Government websites. The parallel corpus covers many domains: the Bible, other literature, Wiki data relating to many topics, Government policies, and general conversation. We have processed the raw data collected from the books, websites, performed sentence alignments (a mix of manual and automatic alignments) and released the corpus in a form suitable for various NLP tasks. Corpus Format ------------- OdiEnCorp 2.0 is stored in simple tab-delimited plain text files, each with three tab-delimited columns: - a coarse indication of the domain - the English sentence - the corresponding Odia sentence The corpus is shuffled at the level of sentence pairs. The coarse domains are: books ... prose text dict ... dictionaries and phrasebooks govt ... partially formal text odiencorp10 ... OdiEnCorp 1.0 (mix of domains) pmindia ... PMIndia (the original corpus) wikipedia ... sentences and phrases from Wikipedia Data Statistics --------------- The statistics of the current release are given below. Note that the statistics differ from those reported in the paper due to deduplication at the level of sentence pairs. The deduplication was performed within each of the dev set, test set and training set and taking the coarse domain indication into account. It is still possible that the same sentence pair appears more than once within the same set (dev/test/train) if it came from different domains, and it is also possible that a sentence pair appears in several sets (dev/test/train). Parallel Corpus Statistics -------------------------- Dev Dev Dev Test Test Test Train Train Train Sents # EN # OD Sents # EN # OD Sents # EN # OD books 3523 42011 36723 3895 52808 45383 3129 40461 35300 dict 3342 14580 13838 3437 14807 14110 5900 21591 20246 govt - - - - - - 761 15227 13132 odiencorp10 947 21905 19509 1259 28473 24350 26963 704114 602005 pmindia 3836 70282 61099 3836 68695 59876 30687 551657 486636 wikipedia 1896 9388 9385 1917 21381 20951 1930 7087 7122 Total 13544 158166 140554 14344 186164 164670 69370 1340137 1164441 "Sents" are the counts of the sentence pairs in the given set (dev/test/train) and domain (books/dict/...). "# EN" and "# OD" are approximate counts of words (simply space-delimited, without tokenization) in English and Odia The total number of sentence pairs (lines) is 13544+14344+69370=97258. Ignoring the set and domain and deduplicating again, this number drops to 94857. Citation -------- If you use this corpus, please cite the following paper: @inproceedings{parida2020odiencorp, title={OdiEnCorp 2.0: Odia-English Parallel Corpus for Machine Translation}, author={Parida, Shantipriya and Dash, Satya Ranjan and Bojar, Ond{\v{r}}ej and Motlicek, Petr and Pattnaik, Priyanka and Mallick, Debasish Kumar}, booktitle={Proceedings of the WILDRE5--5th Workshop on Indian Language Data: Resources and Evaluation}, pages={14--19}, year={2020} }

This thesis on the Swedish voluntary women’s defence organisation Riksförbundet Sveriges lottakårer (SLK), studies how the board and the members of SLK discussed the problems and the meaning of voluntary defence work during the 1960s and 1970s. Furthermore the thesis aims to engage with previous research and give an explanation to why people may be interested in voluntary defence. Views varied greatly within SLK: some meant that the voluntary grounds of participation gave a sense of elite status to the organisation; however others saw voluntary work as untenable, since more women at the time had both work and family obligations. Another issue was professionalisation versus socialisation. While some members wanted to see a centralisation and professionalisation of the organisation, others found it more important to safeguard the social functions and independence of the local corps unit. The results suggest that gender structures as well as individual agency create multiple reasons and meanings behind voluntary defence work.

The collective monograph Memory of Central and Eastern Europe: Past Traumas, Present Challenges, Future Horizons provides a unique platform for collaboration of different theoretical and conceptual approaches to memory and related studies. The first part of the book focuses on the issues related to the problematic aspects of memory in the context of urban and media space. The second part deals with the interaction between nation, national history and memory. The third part is devoted to the representation of memory and sheds light on the specific role of text as a means of cultural memory.

Digitization is an increasingly important subject within the archival profession, especially as our society grows more and more dependent on digital assets. People generally expect everything to be available to them online no matter the time or place, otherwise the material might as well not exist. When it comes to preserving and making our cultural heritage available through digitization this poses an issue. Digitization takes time and requires resources, which forces the archives to decide which material to digitize, on what grounds and how to best make the digital material accessible. This essay explores the digitization of our cultural heritage through the lens of value theory. By examining what type of materials the archives prioritize for digitization, why, and how they utilize the digitized material we can observe the values and principles these approaches stem from. The study is conducted by using content analysis to examine documents such as annual reports, budget documentation and digitization strategies from two archival organizations during the time periods 2010–2015 and 2020–2024. The results show that the choices the archives make regarding what to digitize are based largely on preservation, demand and research requests. Both organizations also view digitization as a key element to working in alignment with democratic values. The approaches to this varied however as one organization focused more on streamlining their organization through digitization, while the other focused more on accessibility and spreading knowledge through their digitized materials.