Update README with latest bm25 metrics and tokenization strategies

README.md

@@ -433,7 +433,7 @@ A lightweight, evaluation-ready subset of [MIRACL](https://huggingface.co/datase
 
 | Model | avg | ar | bn | de | en | es | fa | fi | fr | hi | id | ja | ko | ru | sw | te | th | yo | zh |
 |---|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|
-| bm25 | 0.4891 | 0.4745 | 0.4760 | 0.4234 | 0.4913 | 0.4712 | 0.4837 | 0.5726 | 0.3915 | 0.5988 | 0.5178 | 0.5569 | 0.4627 | 0.4379 | 0.4472 | 0.5884 | 0.5798 | 0.4664 | 0.3632 |
+| bm25 | 0.5138 | 0.5363 | 0.4760 | 0.4724 | 0.5831 | 0.5582 | 0.4837 | 0.6503 | 0.4241 | 0.5988 | 0.5178 | 0.5569 | 0.4627 | 0.4829 | 0.4472 | 0.5884 | 0.5798 | 0.4664 | 0.3632 |
 | e5-small | 0.7181 | 0.7463 | 0.7127 | 0.6579 | 0.6968 | 0.7278 | 0.7203 | 0.7808 | 0.6339 | 0.6612 | 0.6261 | 0.7044 | 0.6730 | 0.7271 | 0.7025 | 0.9779 | 0.8896 | 0.6130 | 0.6737 |
 | e5-large | 0.7791 | 0.8186 | 0.7920 | 0.7925 | 0.7386 | 0.8026 | 0.7522 | 0.8347 | 0.6710 | 0.7586 | 0.6229 | 0.7449 | 0.6907 | 0.7927 | 0.7540 | 0.9779 | 0.9174 | 0.8168 | 0.7465 |
 | bge-m3 | 0.7873 | 0.8193 | 0.8417 | 0.7337 | 0.7406 | 0.7419 | 0.7733 | 0.7977 | 0.7355 | 0.7402 | 0.6636 | 0.7975 | 0.7266 | 0.7993 | 0.7635 | 0.9926 | 0.9090 | 0.8547 | 0.7402 |
@@ -445,33 +445,30 @@ A lightweight, evaluation-ready subset of [MIRACL](https://huggingface.co/datase
 
 ## BM25 tokenization strategy
 
-- Auto selection chooses the best strategy per split using retrieval metrics.
-- Strategy types:
-  - `transformer`: `Qwen/Qwen3-0.6B` tokenizer
-  - `wordseg`: language-specific word segmentation (`ja`, `zh`, `th`, `ko`)
-  - `stemmer`: `PyStemmer`
-  - `whitespace`: `str.split()`
+- `bm25` was generated with `--auto-select-best-splitter`.
+- Candidate strategies: `transformer`, `stemmer`, `wordseg`, `nltk_stem`, `nltk_stem_stop`, `english_regex`, `english_porter`, `english_porter_stop`, `whitespace`.
+- Selection metric: best `nDCG@100` per split.
 
 Selected strategy by split:
 
 | Split | Strategy | Details |
 |---|---|---|
-| `ar` | `stemmer` | PyStemmer (`arabic`) |
+| `ar` | `nltk_stem` | NLTK stemmer |
 | `bn` | `whitespace` | `str.split()` |
-| `de` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
-| `en` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
-| `es` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
+| `de` | `nltk_stem_stop` | NLTK stemmer + stopword removal |
+| `en` | `english_regex` | regex tokenization (no stemming) |
+| `es` | `nltk_stem` | NLTK stemmer |
 | `fa` | `whitespace` | `str.split()` |
-| `fi` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
-| `fr` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
+| `fi` | `nltk_stem_stop` | NLTK stemmer + stopword removal |
+| `fr` | `nltk_stem_stop` | NLTK stemmer + stopword removal |
 | `hi` | `stemmer` | PyStemmer (`hindi`) |
 | `id` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
 | `ja` | `wordseg` | ja (fugashi + unidic-lite) |
 | `ko` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
-| `ru` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
+| `ru` | `nltk_stem` | NLTK stemmer |
 | `sw` | `whitespace` | `str.split()` |
 | `te` | `whitespace` | `str.split()` |
-| `th` | `wordseg` | th (pythainlp (newmm)) |
+| `th` | `wordseg` | th (pythainlp newmm) |
 | `yo` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
 | `zh` | `transformer` | tokenizer: `Qwen/Qwen3-0.6B` |
 
@@ -482,4 +479,4 @@ Selected strategy by split:
 
 ## License
 
-Other. This dataset is derived from MIRACL and follows upstream licensing and attribution requirements.
+Other. This dataset is derived from MIRACL and follows upstream licensing and attribution requirements.