Download the PHP package ali-apadana/persian-keyword without Composer
On this page you can find all versions of the php package ali-apadana/persian-keyword. It is possible to download/install these versions without Composer. Possible dependencies are resolved automatically.
Download ali-apadana/persian-keyword
More information about ali-apadana/persian-keyword
Files in ali-apadana/persian-keyword
Package persian-keyword
Short Description A Laravel-first toolkit for extracting keywords, phrases, and entities from Persian text.
License MIT
FAQ
Example:
In general, it is recommended to use always a project to download your libraries. In an application normally there is more than one library needed.
- Some hosting areas are not accessible by a terminal or SSH. Then it is not possible to use Composer.
- To use Composer is sometimes complicated. Especially for beginners.
- Composer needs much resources. Sometimes they are not available on a simple webspace.
- If you are using private repositories you don't need to share your credentials. You can set up everything on our site and then you provide a simple download link to your team member.
- Simplify your Composer build process. Use our own command line tool to download the vendor folder as binary. This makes your build process faster and you don't need to expose your credentials for private repositories.
Informations about the package persian-keyword
PersianKeyword
ali-apadana/persian-keyword is a Laravel-first package for extracting keywords, important phrases, and dictionary-based named entities from Persian (Farsi) text.
Version 2.4 provides native phrase-aware ranking, person and organization recognition, and sport-event recognition. It remains fully PHP/Laravel based and requires no external service.
HTML content is supported: tags are removed and entities such as ‌ and are decoded before extraction.
Known letter-and-number model names remain one token. For example, اف ۳۵ and F-35 are extracted as complete model identifiers rather than separate words.
The built-in stop-word dictionary includes broad Persian coverage: formal connectors, pronoun variants, number and temporal forms, news boilerplate, auxiliary verbs, and frequent non-topical verb forms. Pattern rules also cover inflections such as می‌تواند, نمی‌توانند, and می‌گوید. You can point resources.stopwords at your own PHP list after publishing the configuration if your content needs different rules.
Common past and perfect verb forms such as گرفته‌اند, کرده‌ایم, and رسیدند are filtered with native patterns, so sentence-ending verbs do not become keywords.
Requirements
- PHP 8.2 or later
- Laravel 12
Installation
Laravel discovers the package automatically. To customize its defaults, publish the configuration file:
Usage
The extractor ranks meaningful phrases alongside single terms. For example, a title containing حمله سایبری به سیستم‌های آبرسانی ۱۲ ایالت آمریکا prioritizes حمله سایبری, آمریکا, and آبرسانی instead of standalone numbers or generic heads such as سیستم.
It also recognizes organization phrases without requiring a dictionary entry. For example, باشگاه استقلال is detected as an organization entity and receives a ranking boost.
Names that immediately precede an official title are detected as people. For example, محمد باقر قالیباف رئیس مجلس detects محمد باقر قالیباف as a person; government organizations such as مجلس شورای اسلامی are also recognized as complete entities.
The same native rule supports cultural and academic roles: علی قائمی is recognized as a person in a phrase such as علی قائمی، اندیشمند مازندرانی.
The native facility rules recognize numbered South Pars refineries such as پالایشگاه سوم پارس جنوبی. The built-in entity dictionary also includes the Industrial Development and Renovation Organization of Iran and its acronym, ایدرو.
Common sport events such as جام ملت‌های آسیا, جام جهانی, لیگ برتر, and بازی‌های المپیک are also recognized as event entities and receive an additional ranking boost.
Sports support also recognizes تیم ملی فوتبال ایران, club-country names such as ویتبسک بلاروس, and disciplinary events such as اخراج.
For general news and health content, native rules recognize events such as تیراندازی در مدرسه and terms such as سرطان مری; long institutions such as دانشگاه علوم پزشکی تهران are also detected.
Pass options for a single extraction when needed:
keywords() returns the ranked keyword strings. Use keywordScores() when you also need each keyword's numeric score and frequency:
Each detected entity records its name, type, and whether it came from the title or body:
The 0.2 result also exposes the normalized source text in $result->meta():
By default, tokens() removes the package stop words. Preserve them for a specific call when you need the raw token stream:
phrases() returns unique two- and three-term candidates, such as آموزش لاراول and برنامه نویسی لاراول. A phrase never spans a stop word.
You can also use the service directly:
Configuration
The initial dictionary is stored in resources/entities.php. After publishing the package configuration, point resources.entities to your own PHP dictionary file when your application needs additional entity types or names.
Development
Notes
Entity detection in 1.0 is dictionary-based. It is fast and deterministic, but it does not infer entities that are absent from the configured dictionary.
License
Released under the MIT License.
