#nlp

Japanese-English dictionary data via the open Jisho.org API (no key). The search endpoint queries the dictionary for words and kanji compounds and accepts English, romaji, kana or kanji as input; each entry carries its Japanese writings (word + reading), English senses with parts of speech and usage tags, JLPT level and a common-word flag, with an optional filter for common words only. The word endpoint returns the single best — preferably common — match for a keyword, ideal for quick look-ups and language tools. Real dictionary data straight from Jisho, cached briefly for speed — no key. 3 endpoints. Ideal for language-learning apps, furigana and reading helpers, vocabulary tools and Japanese NLP enrichment.

api.oanor.com/jisho-api

Generate n-grams from text, with frequency counts — entirely locally. The ngrams endpoint breaks text into contiguous sequences of n tokens and returns each distinct n-gram with how often it occurs, ranked by frequency: word n-grams (unigrams, bigrams, trigrams and beyond) for phrase and collocation analysis, or character n-grams (shingles) for fuzzy matching, language detection and indexing. The range endpoint produces every size from a minimum to a maximum in a single call (for example 1–3 grams), which is exactly what you need to build feature vectors. Choose word or character mode, whether to lower-case first, and a top-N limit to keep only the most frequent. Word tokenization is Unicode-aware and keeps internal apostrophes and hyphens (don't, well-known) as single tokens. Everything runs locally and deterministically, so it is fast and private. Ideal for text mining and NLP feature extraction, language modelling and autocomplete, search indexing and shingling, plagiarism and similarity detection, and keyword and collocation analysis. Pure local computation — no key, no third-party service, instant. Live, nothing stored. 3 endpoints. This produces n-grams and counts; for extractive summaries and keywords use a summarize API and for grapheme/character counting use a text-segmentation API.

api.oanor.com/ngram-api

Summarize text and pull out its keywords — no AI key, no external model. The summarize endpoint is extractive: it scores every sentence by word frequency and position and returns the most representative ones (ask for a fixed number of sentences or a fraction of the original), keeping the author's exact wording and order. The keywords endpoint ranks the most salient terms with their counts and a relative score, filtering out stopwords. Because it is deterministic and runs locally, the same text always gives the same result, instantly and privately. Perfect for article previews and TL;DRs, search snippets, tagging and content triage, and feeding shorter context to downstream tools. Pure local computation — no third-party service; send long text via POST. Live, nothing stored. 3 endpoints. Distinct from sentiment/NLP analysis, stopword lists and Unicode text segmentation.

api.oanor.com/summarize-api

Reduce words to their linguistic root (stem) with the classic Snowball stemming algorithms — running → run, fishing → fish, nationalization → nation — across 24 languages including English, German, French, Spanish, Italian, Portuguese, Dutch, Russian, Arabic, Finnish, Swedish and more. Stem a whole text (every word, returning both the per-word mapping and the fully stemmed text) or a single word. Stemming is the core normalisation step behind search engines, query expansion, text indexing, keyword matching and NLP preprocessing. Pure local computation — no key, no third-party service, instant. Live, nothing stored. 4 endpoints. Distinct from sentiment/NLP analysis and fuzzy string matching.

api.oanor.com/stemmer-api

Score how easy a piece of text is to read using the standard, peer-reviewed readability formulas — Flesch Reading Ease, Flesch-Kincaid Grade, Gunning Fog, SMOG, Coleman-Liau and the Automated Readability Index. Pass text and get all six scores back together with the underlying counts (words, sentences, syllables, complex and polysyllabic words, letters and characters), an averaged grade level, an estimated reading time and a plain-English interpretation of the reading ease. A second endpoint counts syllables for a word or for every word in a phrase. Supply text inline via ?text=, as a query parameter or in a request body; everything is computed locally with no network calls, so it is fast and deterministic. Built for content and copywriting tools, SEO and editorial workflows, education and accessibility (plain-language) checks, and UX-writing review. A readability scorer — distinct from sentiment/NLP analysis (nlp), spelling and grammar checking (grammar), the case and text utilities (text) and string similarity (similarity). No upstream key, no cache.

api.oanor.com/readability-api

The Hugging Face Hub as an API — the central, open registry of machine-learning models and datasets that powers much of the modern AI ecosystem. This API wraps the public huggingface.co Hub into clean JSON. /v1/models searches the Hub's models and lets you filter by task (pipeline_tag — e.g. text-generation, text-to-image, image-classification, automatic-speech-recognition, sentence-similarity) and by library (transformers, diffusers, sentence-transformers, …), sorted by downloads, likes, last-modified, created or trending score — each model returned with its id, author, task, library, download and like counts, license, tags and timestamps. /v1/model?id=google-bert/bert-base-uncased returns a single model's full metadata. /v1/datasets searches ML datasets the same way, and /v1/dataset?id=ILSVRC/imagenet-1k returns a single dataset's metadata. Ids are in org/name form (take them from the search endpoints). Ideal for ML and MLOps tooling, model-discovery and comparison sites, AI leaderboards and dashboards, and AI assistants that recommend models. Data comes from the public Hugging Face Hub (free to use). This is the AI/ML model and dataset hub — distinct from software-package registries (npm, PyPI, Maven, NuGet) and academic paper indexes (arXiv).

api.oanor.com/huggingface-api

Stopword lists and removal for 58 languages. Fetch the full stopword list for a language, see all supported languages with their word counts, check whether a single word is a stopword, or strip stopwords out of a block of text to get a clean keyword stream. Built on the open stopwords-iso dataset and served entirely in-memory, so responses are instant and the service is always available. Ideal for search indexing and relevance, NLP preprocessing and text mining, keyword extraction, tag generation and content tooling.

api.oanor.com/stopwords-api

Find words by meaning, sound and spelling — similar-meaning words (thesaurus), rhymes, autocomplete suggestions and wildcard spelling matches. Backed by Datamuse. Ideal for writing assistants, autocomplete, crosswords, word games, poetry tools and NLP preprocessing.

api.oanor.com/words-api

Translate text between 40+ languages with a confidence score and alternative suggestions. Simple GET interface, no model hosting — ideal for localising content, chat messages, product data and user-generated text.

api.oanor.com/translate-api

Analyse text in real time: sentiment scoring (positive / negative / neutral with the matched words), automatic language detection across 180+ languages, and a combined analysis endpoint with text statistics. No setup, no model hosting.

api.oanor.com/nlp-api