The Oracle Cloud Infrastructure (OCI) has a set of AI services which include:
- Language (GA)
- Anomaly Detection (GA)
- Speech (LA)
- Vision (LA)
Language and Anomaly are Generally Available and the others are still on a Limited Availability mode, and should be released in the upcoming months.
I will briefly cover the Language AI service in this post.
Language AI
Language is serverless and multi-tenant service that is accessible using REST API calls. It provides pretrained models that are frequently retrained and monitored to provide you with the best results. You can automate sophisticated text analysis at scale without any machine learning expertise. -> from documentation
Find the Service
In your OCI console the Language service is under Analytics & AI
Once you open it, a prefilled random text will help us to immediately test the service.
We can choose which tools to use for the analysis.
The language detection service identifies which language the input text is in. It is able to identify a wide list of languages. Check here the list. It is recommended at least 10 words for better results, which is reflected in the confidence percentage.
The text classification service analyses the text and identifies categories for the content with a confidence score.
Named Entity Recognition (NER) detects named entities in text. The entire list of supported entities is here.
The Key Phrase Extraction as the name indicates, extracts key words from the text. This can be very interesting in the context of automating the routing of emails or messages.
And finally the most interesting one in my opinion. Sentiment Analysis.
Sentiment analysis analyzes the subjective information in an expression. For example, opinions, appraisals, emotions, or attitudes toward a topic, person, or entity. Expressions are classified, with a confidence score, as positive, negative, or neutral.
Lets look at another example:
All of these services can be accessed with the OCI Console (all the above examples), the REST API, SDK’s and OCI CLI.
SDK’s
The API Reference documentation also offer code examples on how to connect to this service. Pretty handy and saves time 🙂
PM Webcast
I just found this webcast available on YouTube. It’s from the PM team and it is a Get Started session on the product and it even shows how to leverage other services like Data Integration and Analytics to create a more robust use case.