Our Work

The study aimed to analyze the transformation of business models within the digital sports ecosystem, with particular emphasis on the role of e-commerce and digital marketing in contemporary sports ma…

The study aimed to analyze the transformation of business models within the digital sports ecosystem, with particular emphasis on the role of e-commerce and digital marketing in contemporary sports management. To this end, a systematic literature review adopted an integrative approach and followed the PRISMA protocol. The bibliographic search covered Scopus, Web of Science, and ScienceDirect for the 2010–2026 period, with an updated search conducted before manuscript completion to incorporate publications from 2026 that met the inclusion criteria. The systematic review analyzed 50 studies selected according to the protocol. In addition, theoretical and methodological literature supported the conceptual framework and research design. The findings reveal a transition that has consolidated over several years, as traditional models increasingly give way to more digital and, in many cases, hybrid configurations. Social platforms, live commerce, data analytics, and immersive experiences all contribute to this reconfiguration. Another prominent finding concerns the interdependence among technological, commercial, and relational dimensions. Rather than evolving along separate paths, these three dimensions intersect and ultimately reshape how the sports sector creates and captures value. The study's main contribution lies in a systemic conceptual framework designed to articulate these dimensions, as the literature predominantly offers fragmented perspectives that examine each aspect independently. The proposed framework provides an analytical tool for examining both the opportunities and risks associated with e-commerce in the sports sector, with particular relevance when considered from a Latin American perspective.

No description provided.

Description Replication package for a study of how to measure comprehension debt, the gap between code entering a system and any recoverable account of why it has the form it does, in software written…

Description Replication package for a study of how to measure comprehension debt, the gap between code entering a system and any recoverable account of why it has the form it does, in software written by autonomous coding agents. The package supports the paper's central claim: that before a context-ablation experiment can be interpreted, the measurement instrument itself has to be validated. Four validity preconditions. The paper proposes that a context-ablation experiment must demonstrate, before any treatment effect is read: that the ablation actually changed the maintainer's input (manipulation fidelity); that the measure responds to a positive control (responsiveness); that a shape-matched sham perturbation does not reproduce the effect (sham-arm invariance); and that test-retest variation is smaller than the equivalence margin (reliability). This package contains the data and code for checking all four. Three designs on one matched pool. All designs draw on 1,001 code units mined from agent- and human-authored pull requests in the AIDev corpus (repositories with more than 100 stars), and each admits a different subset. (1) An observational design correlates an LLM-based comprehension score with 180-day downstream churn over 525 units in 143 repositories; its interval per standard deviation is RR [0.818, 1.387], which lacks the resolution to separate the hypotheses. (2) A within-unit maintenance-replay experiment replays 95 historical repair tasks under three context conditions with two maintainer model families; 81.1% of tasks give the same outcome under every condition, so the pass/fail outcome saturates. (3) A judge-free defect-localization probe, scored as line-level intersection over union against the historical fix, satisfies all four preconditions for the primary model on the 59 leakage-free tasks. Audit record. Four defects were found in the replay instrument after an earlier version of this package had been released with a harness that reproduced every published number: a mining defect that accepted 36 pull requests' own squash commits as their downstream fixes, leaking the answer; a headline model that had not converged; a scoring ladder that let an LLM judge override a measured behavioural disagreement; and a pre-registered judge-validation gate that returns Cohen's kappa = 0.239 (95% CI [0.088, 0.414]) against a 0.60 threshold, with inter-rater agreement among three human raters itself low (Fleiss' kappa = 0.178). CORRECTIONS.md documents each defect with its mechanism, its corrected estimate, and the figures an earlier draft got wrong. This is the package's main methodological point: a reproducible pipeline is not necessarily a valid one, because a harness recomputes a contaminated frame faithfully. Principal findings. Revealing the true fix location raises localization by 0.356 IoU for the primary model and 0.312 for the second, establishing responsiveness. Blanking in-code documentation lowers localization by 0.048 and 0.078 IoU respectively; for the primary model the effect is attributable to what the documentation says rather than to its form or to the visible fact of redaction. These effects are directional and do not consistently survive correction for multiple comparisons. Removing recorded pull-request deliberation produces no effect distinguishable from zero. The study does not establish that agent-authored code carries more comprehension debt than human-authored code; that contrast remains descriptive and inconsistent across model families. Contents. Mining and experiment code, the mined frames, all scored results, the human rater packets and their verdicts, the pre-specifications written before the final run, and a verification harness that regenerates every number in the paper from committed data without calling a model or touching the network. Committed harness outputs are included so figures can be checked without executing anything. The offline test suite is 118 tests. Both the original 95-task frame and the corrected 61-task frame are retained so the contamination described in the paper stays reproducible rather than being quietly repaired.

This dataset contains cropped and pseudo de-identified videos. All personal metadata, or other identifiable information have been removed. Access is provided via Zenodo under restricted permissions an…

This dataset contains cropped and pseudo de-identified videos. All personal metadata, or other identifiable information have been removed. Access is provided via Zenodo under restricted permissions and a signed user agreement, ensuring use exclusively for research purposes and the public interest, in accordance with GDPR requirements for scientific research (Articles 4, 5, 6, 9, 32, 89; Recitals 26, 33, 50, 156, 159). Users must not attempt re-identification, redistribute the data, or use it commercially. Acute Vertigo Eye-Movement Video Database – Summary This open-source video database captures eye movements from 100 patients presenting with acute vertigo (AV) in the emergency department (ED) of a UK tertiary hospital. It was designed to support machine learning (ML) model development for automated assessment of nystagmus and subtle ocular motor disorders, as well as for educational and research purposes. Dataset Content Total Videos: 1,333 short Eye-movement videos (5–10 seconds each), in MP4 format. Frames: Over 400,000 RGB images Annotations: Example binary semantic segmentation masks (iris and sclera; PNG format): A small representative subset of binary segmentation masks delineating the iris and sclera is provided to illustrate the segmentation outputs.   Ground Truth (GT): Clinical labels independently verified by three vestibular specialists, including two senior specialists. Includes test results, nystagmus type and direction, and clinical outcomes established after ≥2 months of follow-up. Participant Demographics Total Participants: 100 (57 females, 43 males; aged 18–97) Age Distribution: 18–30: 7% 31–50: 31% 51–70: 43% 71+: 19% Ethnicity Distribution: White: 51% (≈30% White British) Asian: 18% Black: 17% Mixed/Other: 14% Gender & Age Coverage: Broadly balanced, though younger patients slightly underrepresented. Technical Details Videos acquired in real-world Adult ED settings using handheld smartphone-based recording Pseudo de-identified and cropped eye images to ensure GDPR compliance Annotations and segmentations were performed locally using MATLAB™ Image Segmenter Binary semantic segmentation masks provide pixel-level ground truth for ML calibration Database Structure ```Smart-Vertigo Dataset resources│├── ZENODO│   ├── README.md│   ├── 01_Cropped_Videos_No_Audio/│   ├── 02_Smart_Vertigo_Dataset.xlsx│   ├── 03_Binary_Segmentation_Masks/│   └── Data_Use_Agreement_SMART_VERTIGO.docx│└── GITHUB    └── Smart-Vertigo-Dataset/        ├── README.md        └── CODE/            └── video_processing/                ├── README.md                ├── audit_and_remove_video_audio.py                ├── extract_video_frames.py                ├── spatial_crop_320x320.py                └── requirements.txt``` Diversity & Fairness The dataset reflects real-world biological and situational variability, including differences in eye size, iris color, scleral characteristics, and lighting conditions. By encompassing demographic diversity and environmental variability, it may support the development of robust, generalizable, and equitable ML models. Access & Compliance Available on Zenodo under restricted access, subject to a signed Data Use Agreement (DUA). GDPR-compliant for scientific research conducted in the public interest

Mazkur maqolada bank tizimida ekologik, ijtimoiy va korporativ boshqaruv (ESG) omillarining samaradorligini oshirish bo'yicha xorijiy mamlakatlar tajribasi tizimli va qiyosiy jihatdan tahlil qilinadi.…

Mazkur maqolada bank tizimida ekologik, ijtimoiy va korporativ boshqaruv (ESG) omillarining samaradorligini oshirish bo'yicha xorijiy mamlakatlar tajribasi tizimli va qiyosiy jihatdan tahlil qilinadi. Tadqiqotda ESG tamoyillarining banklarning barqaror rivojlanishi, kredit va investitsiya qarorlarini takomillashtirish, iqlim va boshqa noan'anaviy moliyaviy xatarlarni boshqarish hamda korporativ boshqaruv sifatini oshirishdagi o'rni o'rganiladi. Yevropa Ittifoqi mamlakatlari, Buyuk Britaniya, AQSh va Osiyo davlatlarida ESG omillarini bank faoliyatiga integratsiya qilishning institutsional mexanizmlari, prudensial talablar, barqaror moliyalashtirish vositalari va axborotni oshkor qilish amaliyotlari qiyosiy tahlil qilinadi. Shuningdek, xorijiy tajribaning O'zbekiston bank tizimi sharoitida qo'llash imkoniyatlari, mavjud institutsional cheklovlar va ESG samaradorligini oshirishning ustuvor yo'nalishlari asoslanadi. Tadqiqot natijalari bank sektorida barqaror moliyalashtirishni kengaytirish, ESG mezonlarini risk-menejment tizimiga integratsiya qilish va banklarning uzoq muddatli moliyaviy barqarorligini mustahkamlashga qaratilgan ilmiy-amaliy takliflarni ishlab chiqishga xizmat qiladi.

Analysis code and intermediate datasets for Zhu, Liu, Kelp, and Naeger (2026), "Capturing Time-Resolved Prescribed Fire Emissions with TEMPO Special Observations", Geophysical Research Letters (manusc…

Analysis code and intermediate datasets for Zhu, Liu, Kelp, and Naeger (2026), "Capturing Time-Resolved Prescribed Fire Emissions with TEMPO Special Observations", Geophysical Research Letters (manuscript 2026GL125411). tempo-rx-fires-code-v1.0.zip contains the tempo_fires Python package (TEMPO Level-2 NO2 extraction, HRRR wind regridding, wind-aware plume detection, cross-wind line-density integration, EMG fitting, quality control, uncertainty budget, GOES/HMS/FIRMS overlays, and the top-down versus bottom-up comparison) together with the scripts used for every figure and table (MIT License). tempo-rx-fires-data-outputs-v1.0.zip contains the per-scan wind-aligned analysis windows, line densities, and EMG fit results for the QC-passing TEMPO scans of the four prescribed fires (Geneva State Forest, AL, 27 March 2025; Eglin-Blackwater, FL, 28 March 2025; Chattahoochee-Oconee National Forest, GA, 8 April 2026; Fort Stewart, GA, 12 April 2026), the coincident TROPOMI fits, and the CSV tables behind Tables 1, S1, S2, and S3 (CC BY 4.0). tempo-rx-fires-data-inputs-v1.0.zip contains the 5-minute GOES-16/18/19 ABI fire radiative power per burn unit, HMS smoke polygons, MODIS/VIIRS active-fire detections, digitized burn-unit polygons, bottom-up emission tables, and TROPOMI NO2 regridded to the TEMPO grid (CC BY 4.0). Raw TEMPO Level-2 V04 granules (NASA ASDC, doi:10.5067/IS-40e/TEMPO/NO2_L2.004), HRRR analyses, and RAVE v2 files are not redistributed; the extraction and download scripts are included. See README.md inside each archive for reproduction steps.

Automatize release based on a commit convention

performance: 2x faster extraction (#202) breaking: CLI defaults to extensive search, -f/--fast enables fast mode (#193) breaking: internal helpers changed, regex constants moved from extractors to cor…

performance: 2x faster extraction (#202) breaking: CLI defaults to extensive search, -f/--fast enables fast mode (#193) breaking: internal helpers changed, regex constants moved from extractors to core, filter_ymd_candidate and plausible_year_filter - signatures, reference comparison with datetime | None instead of timestamps (#193, #203) fix: keep time and time zone of regex matches (#194) fix: more robust date parsing and selection, e.g. month names in any case, unpadded dates, mixed separators, impossible dates, outputformat for copyright years (#193, #203) maintenance: code hardening and simplification (#191, #193), require dateparser >= 1.4.2 (#199) maintenance: support Python 3.15 tests: full line coverage, leaner test suite docs: wording fixes (#192) New Contributors @apoorvdarshan made their first contribution in https://github.com/adbar/htmldate/pull/194 @andrewkernel made their first contribution in https://github.com/adbar/htmldate/pull/192 @dotlambda made their first contribution in https://github.com/adbar/htmldate/pull/199

Added bagpuss.mdc: config-driven, sharded generation of large catalogues and injection sets (sky-tiled galaxy catalogue, time-sliced injections, zarr store, HTCondor DAG rendering) -- see docs/mdc.rst…

Added bagpuss.mdc: config-driven, sharded generation of large catalogues and injection sets (sky-tiled galaxy catalogue, time-sliced injections, zarr store, HTCondor DAG rendering) -- see docs/mdc.rst. bagpuss mdc detect-injections / assemble-detections: network SNRs for every injection (per-detector duty cycles, via minke), optional frames, asimov event blueprints for the detectable events, and a BAYESTAR skymap per event (write_skymaps). Needs the detection extra. estimation.trigger_eta_max: cap the simple-pe trigger's symmetric mass ratio, which fails at the equal-mass limit; the injected truth is kept as trigger truth. bagpuss mdc export-release / verify-release: assemble the sharded store into flat, checksummed release files (HDF5 catalogue and injection tables, events, skymaps, README, Zenodo metadata). GLADE+-style catalogue export (bagpuss mdc export-glade). Merger-rate models (MergerRate, ConstantMergerRate) with expected_n_mergers / sample_merger_redshifts, and rate-based injection counts in build_injection_set. Observed-galaxy counting and sampling (expected_n_observed, sample_observed_redshifts), and sample_brighter_than, differential_comoving_volume and z_max on luminosity models and universes. Changed Breaking: create_injection_set now takes universe, catalogue, selection, population and n_draw (previously catalogue, population, cosmology, n_draw). Host galaxies are assigned in two steps: a trial host from the universe, then a catalogue row if the survey completeness says the host would be catalogued (host_galaxy_index == -1 otherwise); it is no longer drawn uniformly from the catalogue. docs/injection.rst is updated. Fixed sample_host_galaxies no longer fails on an empty catalogue. MDCConfig.trigger_eta_max is validated to lie in (0, 0.25]. export-release / verify-release: the catalogue is written tile by tile (no full in-memory concatenation); events.yaml is checked against the detectable events; a damaged skymaps tar or HDF5 file is reported rather than raised; the README resolves the default observation window. BAYESTAR skymaps use the configured low-frequency cutoff throughout. Notes The detection stage imports minke.duty_cycle, which is on minke's main branch but not in a release yet (2.2.1 lacks it). The detection extra can only say minke>=2.2.1 until such a release exists; the detection backend raises a clear ImportError if duty_cycle is missing.