This is an automated email from the ASF dual-hosted git repository.
aradzinski pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-nlpcraft-website.git
The following commit(s) were added to refs/heads/master by this push:
new 8c60b58 WIP.
8c60b58 is described below
commit 8c60b5888d5bd7748a39f38a547cba6877bddf28
Author: Aaron Radzinzski <[email protected]>
AuthorDate: Sun Jan 17 22:54:25 2021 -0800
WIP.
---
blogs/how_to_find_something_in_the_text.html | 87 ++++++++++++++++++++++++++++
integrations.html | 2 +-
2 files changed, 88 insertions(+), 1 deletion(-)
diff --git a/blogs/how_to_find_something_in_the_text.html
b/blogs/how_to_find_something_in_the_text.html
index 6ea3db8..04db768 100644
--- a/blogs/how_to_find_something_in_the_text.html
+++ b/blogs/how_to_find_something_in_the_text.html
@@ -89,8 +89,95 @@ publish_date: January 20, 2021
It appears that the main usage pattern of Apache OpenNLP is to build
and train your own models from scratch.
</p>
<h3 class="section-sub-title">Stanford NLP</h3>
+ <p>
+ <img class="img-title" src="/images/corenlp-logo.png" height="64px"
alt="">
+ </p>
+ <p>
+ <a target="_parent" href="https://nlp.stanford.edu/";>Stanford NLP</a>
is a popular and actively developed, mature NLP library that provides a wide
range of
+ functionality. For English it supports the following named entities:
person, location, organization,
+ misc, money, number, ordinal, percent, date, time, duration, set.
Furthermore, built in regular expressions
+ based NER component allows to recognize the following additional named
entities: email, url, city,
+ state_or_province, country, nationality, religion, (job) title,
ideology, criminal_charge, cause_of_death,
+ handle. More information <a target="_blank"
href="https://stanfordnlp.github.io/CoreNLP/ner.html#description";>here</a>.
+ </p>
+ <p>
+ There’s a limited support for German, Spanish and Mandarin languages.
<a target="_blank" href="https://corenlp.run/";>Live demo</a> allows you to test
out
+ various capabilities of Stanford NLP.
+ </p>
+ <p>
+ Stanford NLP is a Java library. Models are available in Maven along
with the project itself.
+ I could not find a detailed description of NER components for
languages other than English. <a target=_blank
href="https://medium.com/sicara/train-ner-model-with-nltk-stanford-tagger-english-french-german-6d90573a9486";>Here</a>
+ and <a target=_blank
href="https://medium.com/@klintcho/training-a-swedish-ner-model-for-stanford-corenlp-part-2-20a0cfd801dd#.vnow3swam";>here</a>
you can find instructions on how to train your own NER components for other
languages.
+ </p>
+ <p>
+ <b>Pros:</b><br/>
+ Maturity of the project. Live and actively developed project with very
good recognition quality
+ (I use the word “good” very subjectively as we won’t go into formal
qualitative metrics of each
+ project here).
+ </p>
+ <p>
+ <b>Cons:</b><br/>
+ The biggest gripe is the usage of <a target="_blank"
href="https://www.wikiwand.com/ru/GNU_General_Public_License";>GNU GPL</a>
license which is all but shun away these days due its viral
+ nature and business unfriendliness. In other words - it is not free
and you have to buy a commercial
+ license if you intend to use it in any serious way. Documentation is
adequate at best and can be a
+ frustrating experience (just like most other academically driven
software projects).
+ </p>
<h3 class="section-sub-title">Google Language API</h3>
+ <img class="img-title" src="/images/google-cloud-logo-small.png"
height="56px" alt="">
+ <p>
+ <a target="_blank"
href="https://cloud.google.com/natural-language";>Google Language API</a>
supports the
+ following named entities for the English language: person, location,
organization, event, work_of_art,
+ consumer_good, other, phone_number, address, date, number, price.
+ </p>
+ <p>
+ Google Language API is available as REST API with the native client
libraries for Java, C#, Python, Go, etc.
+ </p>
+ <p>
+ <b>Pros:</b><br/>
+ Large set of NER components from a trusted NLP-based company like
Google. Scalability and availability of
+ modern SaaS platform developed by Google...
+ </p>
+ <p>
+ <b>Cons:</b><br/>
+ REST API inherently limits the performance of the final solution -
making it almost impossible to be used
+ in any “real-time” applications. Free only for a small number of
transactions, paid after that. Not open source.
+ </p>
<h3 class="section-sub-title">spacy</h3>
+ <img id="spacy" class="img-title" src="/images/spacy-logo.png"
height="48px" alt="">
+ <p>
+ <a target="_blank" href="https://spacy.io";>spaCy</a> is a Python
library that provides one of the best, if not the best, collection of NER
components.
+ <a target="_blank"
href="https://spacy.io/api/annotation#named-entities";>Here</a> you can see a
full list of supported NERs.
+ </p>
+ <p>
+ <b>Pros:</b><br/>
+ Actively developed and mature project. Open source with MIT license.
One of the best
+ documentation among similar projects. One of the most popular NLP
libraries among a few dozens of available
+ libraries for the Python community.
+ </p>
+ <p>
+ <b>Cons:</b><br/>
+ Python - which is rarely used for production level applications. Slow,
often unacceptably slow,
+ performance (due to Python as well). Lack of 1st grade support for
language other than English.
+ </p>
+</section>
+<section>
+ <h2 class="section-title">Additional Capabilities of Apache NLPCraft</h2>
+ <p>
+ Let’s take a look at what Apache NLPCraft brings different or
additionally to the table.
+ </p>
+ <p>
+ When it comes to NER components, Apache NLPCraft provides the
following:
+ </p>
+ <ul>
+ <li>Built-in NER components for date, geographical locations,
numerics, sorting, limiting, and few others with all of them supporting the
extraction of the normalized values and extensive metadata.</li>
+ <li>Integration with external NER components from Apache OpenNLP,
Stanford NLP, Google Language API and spacy.</li>
+ <li>Support for “composable entities” where users can create new
detectable named entities out of existing ones.</li>
+ </ul>
+ <p>
+ While built-in NER components and integration with 3rd party ones is
rather a “pedestrian”
+ capabilities (and you can read about them <a
href="/integrations.html">here</a>) - the “composable entities” is something
that is unique for Apache NLPCraft.
+ Let’s look at it in more detail.
+ </p>
</section>
diff --git a/integrations.html b/integrations.html
index a5d7567..0c589a5 100644
--- a/integrations.html
+++ b/integrations.html
@@ -552,7 +552,7 @@ id: integrations
<ul>
<li>
See Stanford CoreNLP Named Entity Recognition
- <a target="google"
href="https://stanfordnlp.github.io/CoreNLP/ner.html";>documentation</a>
+ <a target="_blank"
href="https://stanfordnlp.github.io/CoreNLP/ner.html";>documentation</a>
for more details on supported token types.
</li>
<li>
