Xetun Bavori
Xetun Bavori NLP Workshop Platform
Enrolling now

Archive — Xetun Bavori

Published Workshop Materials

A structured index of workshop sessions, exercises, and reference texts published through Xetun Bavori on natural language processing. Each entry reflects real classroom activity, not curated reading lists.

Natural language processing workshop session in progress

All Published Entries

Tokenisation

Subword Segmentation and BPE in Practice

A hands-on session covering byte-pair encoding applied to Ukrainian and English corpora. Participants built a vocabulary from scratch using a 40,000-token dataset.

Workshop · 2024

Syntax Parsing

Dependency Trees Without a Framework

Step-by-step construction of a dependency parser using shift-reduce transitions. The exercise used real news sentences rather than textbook examples.

Exercise · 2024

Embeddings

Measuring Semantic Distance Across Domains

Participants compared cosine similarity outputs across four embedding models on the same sentence pairs. Differences between medical and legal corpora were measurable and significant.

Reference · 2023

Sequence Labelling

Named Entity Recognition on Low-Resource Text

An applied session focused on annotating named entities in three under-resourced language varieties. The group worked with raw court records rather than pre-cleaned input.

Workshop · 2023

Language Modelling

Perplexity as a Diagnostic Signal

The session examined perplexity scores as an indicator of model fit across training domains. Participants ran evaluations on n-gram and transformer models side by side.

Reference · 2022

How Each Entry Is Structured

Archive entries follow a fixed format so that returning participants can locate the relevant material without re-reading entire documents.

  1. Step 01

    Topic and category label

    Each entry carries a single NLP topic tag drawn from a fixed taxonomy: tokenisation, parsing, embeddings, sequence labelling, or language modelling.

  2. Step 02

    Session type designation

    Entries are marked as Workshop, Exercise, or Reference. Workshops contain group tasks; Exercises are individual; Reference entries are reading material without tasks.

  3. Step 03

    Year of publication

    Publication year appears in the tag and determines sort order within the year filter in the left column.

  4. Step 04

    Summary description

    A two-to-three sentence description states the specific task, dataset, or analytical question addressed — without restating the title.

Archive by the numbers

The archive spans material produced since 2014. Volume reflects active publishing periods rather than editorial targets.

68 Entries in the archive across all years and session types
5 Core NLP topic areas covered across workshops and exercises
4.9 Average participant rating across reviewed sessions