Research-grade AI pipelines for universities and businesses — patrickchu.net
Extracts structured fields from eleven pinned generative-AI-in-education abstracts, for researchers learning literature-matrix workflows on a small corpus.
Open demo →Tests human-reviewed email triage, reference drafting, and meeting summaries on fictional academic materials, for staff exploring admin workflows safely.
Open demo →Traces a simulated reaction-time experiment from trial data through exclusion rules to a draft Results paragraph, for cognitive researchers testing analysis flow.
Open demo →Walks through back-translation quality checks on eleven bilingual questionnaire items, for researchers testing translation workflow before field work.
Open demo →Re-runs entropy-based classification on one thousand deposited Chinese texts, for readers checking whether translated versus original patterns reproduce.
Open demo →Re-runs acoustic analyses from a Cantonese child speech study and compares group prosody patterns, for readers checking published reproduction claims.
Open demo →Re-runs core analyses from a child-care burnout study on its public longitudinal file, for readers comparing reproduced estimates with published values.
Open demo →Checks a deposited COVID-19 survey against its published analyses, showing which results reproduce and which cannot be verified from the archive.
Open demo →Recalculates evaluation metrics from shared street-image ratings and stored predictions, for readers comparing model scores with survey restorative judgments.
Open demo →Re-runs stroke quality-of-life analyses from a published clinical survey file using open tools, flagging where invariance tests differ from the original.
Open demo →Reproduces pretest-only correlations and regressions from a dyslexia reading study with 35 children, separating statistical match from generalisability claims.
Open demo →Re-runs four COVID-era self-concept studies from one deposited paper, for readers comparing reproduced correlations across US and regional samples.
Open demo →Re-runs two deposited emoji-and-creativity experiments and compares reported mediation results, for readers checking statistical reproducibility.
Open demo →Compares two archived face-perception experiments with the paper's core age estimates, for readers assessing how far the numbers hold up.
Open demo →Compares a deposited vocabulary-learning study's published statistics with an independent re-run, for readers judging reproducibility of word-length effects.
Open demo →Traces a single-subject EEG recording from raw traces to event-related potentials and scalp maps, for students learning standard preprocessing steps.
Open demo →Reproduces classic event-related potential effects from a public multi-paradigm EEG resource, with pre-specified gates and honest amber labels.
Open demo →Re-runs language-attitude comparisons across Cantonese, English, and Mandarin ratings from 540 participants, for readers checking main published effects.
Open demo →Independently codes seventy-seven news articles on inclusive education framing, for media researchers comparing automated labels with published category splits.
Open demo →Retrieves one completed thesis and associated experiment files, then re-runs one verifiable listening study for approximate comparison—not full replication.
Open demo →Reanalyses three deposited reaction-time experiments with a population-averaged model, for readers comparing direction patterns with the published mixed-model results.
Open demo →Compares group-level sentence-comprehension patterns from an open child-language dataset using an approximate model, not individual diagnosis.
Open demo →Recomputes five Provo Corpus eye-movement measures and two reading effects, for readers verifying published word-reading statistics.
Open demo →Applies a fixed five-category codebook to thirty public complaint narratives, for researchers comparing AI labels with existing regulator categories.
Open demo →Compares AI essay scores and feedback with two human raters on twenty-five public middle-school essays, for educators exploring rubric alignment limits.
Open demo →Reproduces regression analyses from a consecutive-interpreting error study with 53 students, for readers checking whether effect directions match the paper.
Open demo →Routes fifteen scripted Cantonese messages through distress flags and crisis handoff rules, for teams reviewing triage safety—not live counselling.
Open demo →Screens sixteen hand-written abstracts against stated inclusion criteria, for review teams testing AI-assisted abstract triage on a pilot set.
Open demo →Shows how a short synthetic lesson moves from audio to coded discourse turns, for educators exploring automated classroom transcription—not real recordings.
Open demo →Maps a synthetic grant-style summary against published funding criteria, for applicants exploring structured pre-submission self-review.
Open demo →Compares two AI-text detection approaches on twenty constructed human-and-machine passage pairs, for readers assessing mechanism limits—not benchmark claims.
Open demo →Tests whether a Cantonese speech pipeline spots planted child pronunciation errors, for clinicians and researchers evaluating mechanism—not diagnosis.
Open demo →Transcribes a synthetic Cantonese passage and extracts word-level pitch and timing, for speech researchers testing prosody measurement on clean audio.
Open demo →Runs exploratory factor analysis on synthetic survey responses with a known three-factor structure, for researchers testing questionnaire workflows.
Open demo →Computes standard reading measures on synthetic Chinese fixation data with known frequency effects, for readers validating measure calculations.
Open demo →Measures turn counts and talk share from a synthetic parent–child Cantonese clip, for researchers testing conversational metrics—not validated field data.
Open demo →Compares rate, accuracy, and expression scores across two synthetic Chinese readings of the same passage, for educators testing fluency measurement—not norms.
Open demo →Re-runs distribution and growth-model checks on a deposited hip-hop collaboration network, for readers comparing patterns with the published analysis.
Open demo →Scores ten Chinese–English translation pairs with planted errors, for teams testing whether an AI judge prefers the polished version and spots flaws.
Open demo →Calibrates sentiment, extracts themes, and drafts summaries on a fixed English review corpus, for analysts learning text-mining with human-check labels.
Open demo →Re-runs corpus comparisons between translated and native English across four genres, for readers checking twenty-four published statistical results.
Open demo →Re-runs deposited eye-tracking analyses comparing translation and paraphrase reading, for readers checking twenty-one published statistical verdicts.
Open demo →Compares glossary-assisted and unassisted translation on fourteen synthetic government-style sentences, for teams testing terminology-memory gains.
Open demo →Compares crowd and machine print age-of-acquisition estimates for about 11,000 English words, for readers evaluating norm coverage and agreement.
Open demo →Processes simultaneous EEG and eye-tracking from natural reading in twelve adults, for researchers exploring underpowered word-frequency effects honestly.
Open demo →No demos match those filters. Clear the filters to see all 45 demos.