• Digital Humanities and Cultural Heritage
• Research dataclose
• Research softwareclose
• Other research productsclose
• Open Access close
• Dataset close
• European Commission close
• EC|H2020 close
• EU close
• FI close

There is widespread interest in the relationship between the neurobiological systems supporting human cognition and emerging computational systems capable of emulating these capacities. Human speech comprehension, poorly understood as a neurobiological process, is an important case in point. Automatic Speech Recognition (ASR) systems with near-human levels of performance are now available, which provide a computationally explicit solution for the recognition of words in continuous speech. This research aims to bridge the gap between speech recognition processes in humans and machines, using novel multivariate techniques to compare incremental ‘machine states’, generated as the ASR analysis progresses over time, to the incremental ‘brain states’, measured using combined electro- and magneto-encephalography (EMEG), generated as the same inputs are heard by human listeners. This direct comparison of dynamic human and machine internal states, as they respond to the same incrementally delivered sensory input, revealed a significant correspondence between neural response patterns in human superior temporal cortex and the structural properties of ASR-derived phonetic models. Spatially coherent patches in human temporal cortex responded selectively to individual phonetic features defined on the basis of machine-extracted regularities in the speech to lexicon mapping process. These results demonstrate the feasibility of relating human and ASR solutions to the problem of speech recognition, and suggest the potential for further studies relating complex neural computations in human speech comprehension to the rapidly evolving ASR systems that address the same problem domain.

The digital text of the 13 books of the "Confessiones" by Augustinus is taken from The Latin Library (http://www.thelatinlibrary.com/august.html). The original text was lemmatized and PoS tagged with the UDPipe tool (using the PROIEL trained model). The output of UDPipe was then checked manually at the CIRCSE Research Centre of the Università Cattolica del Sacro Cuore, Milan, Italy. The linking of the text to the Lemma Bank of the LiLa Knowledge Base was performed at CIRCSE, too.

This dataset contains the dissolved organic matter (DOM) quantification and optical characterisation results from a KOSMOS mesocosm experiment carried out in the framework of the Ocean Artificial Upwelling project. The experiment was carried out in the autumn of 2018 in the oligotrophic waters of Gran Canaria. During the 39 days of experiment nutrient-rich deep water was added to the mesocosms in two modes (singular vs recurring additions), with four levels of intensity. Dissolved organic carbon, nitrogen and phosphorus were quantified with a Shimadzu TOC-5000 and a QuAAtro AutoAnalyzer. The absorption and fluorescence proprieties of DOM were determined making use of an Ocean Optics USB2000+UV-VIS-ES Spectrometer and a Jobin Yvon Horiba Fluoromax-4 spectrofluorometer, respectively. The aim of this dataset was to study the effect of artificial upwelling on the dissolved organic matter pool and its potential implications for carbon sequestration.

ELMo language model (https://github.com/allenai/bilm-tf) used to produce contextual word embeddings, trained on entire Gigafida 2.0 corpus (https://viri.cjvt.si/gigafida/System/Impressum) for 10 epochs. 1,364,064 most common tokens were provided as vocabulary during the training. The model can also infer OOV words, since the neural network input is on the character level.

This dataset contains environmental data (temperature, relative humidity, and, in some cases, light and ultraviolet radiation levels) of partner museums of the European Horizon 2020 CollectionCare project . The following museums provided data to create this compilation and consolidation: Alava Arms Museum (Spain), Alava Fine Arts Museum (Spain), National Historical Museum (Greece), The Ethnographic Open Air Museum of Latvia, The Royal Danish Collection - Rosenborg (Denmark).

We freeze-dried and homogenized 44 samples of c. 0.7-1.8 g dry sediment from core PG1351 covering late glacials and interglacials of MIS 8 to MIS 5e, integrating sediment of 1 cm core depth. Temporal resolution of these samples ranges from 140 to 960 years per sample. For the period between 430 and 405 kyrs ago (end of MIS 12 to MIS 11c), 13 samples of 0.5-1.3 g of dry sediment from ICDP core 5011-1 were taken for MA analyses, integrating sediment of 2 cm core depth. Eight of these 13 samples are from the same core depths as were previously analysed for pollen (Melles et al., 2012). Temporal resolution of these samples varies between 200 and 970 years per sample comparable to core PG1351. Across all samples, temporal resolution is 333 ± 273 years per sample, giving centennial- to millennial scale averages. We extracted the polar lipids of all MA samples using a Dionex Accelerated Solvent Extraction system (ASE 350, ThermoFisher Scientific) at 100°C, 103 bar pressure and two extraction cycles (20 min static time) with 100 % methanol, after an ASE cycle with 100 % dichloromethane. For every sample sequence (n=13-18), we extracted a blank ASE cell and included it in all further steps. We added 60 ng of deuterated levoglucosan (C6H3D7O5; dLVG; Th. Geyer GmbH & Co. KG) as internal standard, and filtered the extract over a PTFE filter using acetonitrile and 5 % HPLC-grade water. We analysed the extracts with an Ultimate 3000 RS ultra-high performance liquid chromatograph (U-HPLC) with thermostated autosampler and column oven coupled to a Q Exactive Plus Orbitrap mass spectrometer (Quadrupole-Orbitrap MS; ThermoFisher Scientific) with heated electrospray injection (HESI) probe at GFZ Potsdam, using measurement conditions adapted from earlier studies (Hopmans et al., 2013;Schreuder et al., 2018;Dietze et al., 2019). Briefly, separation was achieved on two Xbridge BEH amide columns in series (2.1 x 150 mm, 3.5 um particle size) fitted with a 50 mm pre-column of the same material (Waters). The compounds were eluted (flow rate 0.2 mL min-1) with 100 % A for 15 minutes, followed by column cleaning with 100 % B for 15 min, and re-equilibration to starting conditions for 25 min. Eluent A was acetonitrile:water:triethylamine (92.5:7.5:0.01) and eluent B acetonitrile:water:triethylamine (70:30:0.01). HESI settings were as follows: sheath gas (N2) pressure 20 (arbitrary units), auxiliary gas (N2) pressure 3 (arbitrary units), auxiliary gas (N2) temperature of 50 ˚C, spray voltage -2.9 kV (negative ion mode), capillary temperature 300 °C, S-Lens 50 V. Detection was achieved by monitoring m/z 150-200 with a resolution of 280,000 ppm. Targeted data dependent MS2 (normalized collision energy 13 V) was performed on any signal within 10 ppm of m/z 161.0445 (calculated exact mass of deprotonated levoglucosan and its isomers) or m/z 168.0884 (calculated exact mass of deprotonated dLVG) with an isolation window of 0.4 m/z. The detection limit was 2.5 pg on column, based on injections of 0.5 to 5000 pg on column of authentic standards of LVG, MAN, and GAL (Santa Cruz Biotechnology) and dLVG. Integrations were performed on mass chromatograms within 3 ppm mass accuracy and corrected for relative response factors to dLVG (1.08 ± 0.10, 0.76 ± 0.10 and 0.24 ± 0.05 for LVG, MAN, and GAL, respectively), according to known authentic standard mixes injected before and after every measurement sequence and supported by characteristic isomer-specific MS² data. All samples were corrected by subtracting the maximum MA concentrations in the blank duplicates of each ASE sequence. To account for biases due to sediment properties and sedimentation rates, MA influxes (mass accumulation rates in ng cm-2 yr-1) were calculated by multiplying the concentrations (ng g-1) with the sample-specific dry bulk densities (Melles et al., 2007;Wennrich et al., 2016), and the sample's sedimentation rates (cm yr-1) using the age-depth models presented by Nowaczyk et al. (2013) for the the PG1351 and the ICDP-5011-1 cores.

This data set contains relative humidity measurements obtained with sensors installed in the apse vault of the Cathedral of Valencia in Spain. The interest of these sensors is to monitor the conservation conditions of Renaissance frescoes. Included files are: Cathedral_of_Valencia_RH_2008.csv : Relative humidity measurements for the year 2008 Cathedral_of_Valencia_RH_2010.csv : Relative humidity measurements for the year 2010

The dataset contains human similarity ratings for pairs of words. The annotators were presented with contexts that contained both of the words in the pair and the dataset features two different contexts per pair. The words were sourced from the English, Croatian, Finnish and Slovenian versions of the original Simlex dataset.

Audio recording of the study day "Théâtre interculturel et appropriation culturelle. Enjeux éthiques et esthétiques", UQAM, Université du Quebec à Montreal, 10 March 2022.

The clues used to detect the Other component of the Knowledge Source meta-knowledge dimension. (FILE 1 kb)