[WordPress] 外掛分享: Poiskovich – Site Search with Morphology & Relevance

首頁外掛目錄 › Poiskovich – Site Search with Morphology & Relevance
WordPress 外掛 Poiskovich – Site Search with Morphology & Relevance 的封面圖片
全新外掛
安裝啟用
尚無評分
5 天前
最後更新
問題解決
WordPress 6.0+ PHP 7.4+ v0.0.6 上架:2026-08-18

外掛標籤

開發者團隊

⬇ 下載最新版 (v0.0.6) 或搜尋安裝

① 下載 ZIP → 後台「外掛 › 安裝外掛 › 上傳外掛」
② 後台搜尋「Poiskovich – Site Search with Morphology & Relevance」→ 直接安裝(推薦)
📦 歷史版本下載

原文外掛簡介

The built-in WordPress search matches letters, not words. A visitor types “светильники” (lamps) and a page containing “светильник” (lamp) is not found. They type “кровати” (beds) and the product “кровать” (bed) is not found. The visitor concludes you do not sell it and leaves.
Poiskovich reduces words to their stem, so every form of a word is searched as one word.
Russian morphology
Built-in WordPress search matches words literally, so a query for a plural or an inflected form misses posts that use another form of the same word. Poiskovich reduces both the query and the indexed text to word stems, so “светильники” finds “светильник”.
Stemming is enabled out of the box and needs no configuration.
Features

Word forms. “Светильник”, “светильники”, “светильниках” are treated as one word. Works for adjectives and verbs too.
Relevance. A match in the title weighs more than a match in the body. Weights are configurable and any source can be switched off completely.
Weight per post type. Products can be ranked above blog posts with a single number.
WooCommerce products, categories, tags and custom fields — for example an SKU.
Query redirects. Someone searches for “доставка” and lands straight on your shipping page. The rule matches every form of the word.
Query log. A separate tab lists what visitors searched for and did not find — a direct list of what your site is missing or what is named in words your customers do not use.
Highlighting of matched words in the results.
Exclusions — posts that must never appear in search results.

How it works
The plugin builds its own index: posts are split into words in advance instead of being scanned on every visitor request. A LIKE search cannot use database indexes, and on a catalogue of a few thousand products that is noticeable even on good hosting.
The index is compact — one row per “word + post” pair. New and edited posts are indexed automatically; a full rebuild is only needed after changing the settings or after a bulk content edit, and it runs in batches so it does not hit the PHP time limit.
Nothing has to be changed in your theme: the plugin replaces the results of the standard WordPress search, so your theme template, pagination and everything else keep working.
Privacy
The plugin sends nothing anywhere. It makes no external requests and stores all data in your own database. The query log records the search text, the number of results and the date — no IP addresses and no visitor identifiers.

延伸相關外掛

文章
Filter
Mastodon