Download the PHP package dennis-de-swart/php-stanford-nlp-datastore without Composer

PHP Stanford NLP Datastore

Stores NLP data from Stanford CoreNLP server.

What does it do?

It analyses a text using Stanford CoreNLP server, then stores the result.

Which data gets stored?

• OpenIE: these are "Subject-Relation-Object" triples. The concept is similar to "Subject-Verb-Object" triples.

• Named-Entities: if a word is a "Named Entity", like a Location, Name or Time, it will store this data

• Coreference: if there is a reference to a word in another sentence.

How does it work?

• You submit a text.
• The text is analyzed by the Stanford CoreNLP server

• Results are stored in a SQLite file based database. The database file is called "datastore.db"

• The results are displayed on screen
• There is also a search form to find data

This package depends on Stanford CoreNLP Server

This package also depends on PHP-Stanford-CoreNLP-Adapter

Note: since this package contains a full version of the CoreNLP Adapter, you can use all of it's features with this package.

Installation

This package depends on these packages:

Install procedure using the ZIP files

• Install Stanford CoreNLP Server. Check the "php-stanford-corenlp-adapter" package for an installation walkthrough
• Download and unpack the files from this package.
• Copy the files to your to your webserver directory. Usually "htdocs" or "var/www".
• Run a Composer update to install the dependencies

Install as part of another project

• Install Stanford CoreNLP Server. Check the "php-stanford-corenlp-adapter" package for an installation walkthrough

• Add the following lines to your main project's "composer.json" require section:

• Run a Composer update to install the dependencies

• Example code for your main project:

Requirements

• PHP 5.6 or higher: it also works on PHP 7
• Java SE Runtime Enviroment, version 1.8
• Stanford CoreNLP Server 3.7.0
• Windows or Linux/Unix 64-bit OS, 8Gb or more memory recommended.
• Composer for PHP

SQLite Browser

If you need a SQLite browser check here:

Important notes

• Starting the CoreNLP server for the first time, takes some time because it will load a large amount of data.
• After the first startup, the server will be much faster.
• In my experience the Stanford CoreNLP server runs best with 8Gb of memory or more. Start the server with "-mx8g" instead of "-mx4g".
• Also use version 3.7.0 of the server, this gives you the best and quickest results.

Example output

See

• "datastore_result_a.PNG"
• "datastore_result_b.PNG"
• "datastore_result_search.PNG"

and "example.db", this is how a filled database looks like

Any questions?

Let me know. You can create an issue on GitHub. Any bugs will be fixed ASAP.