
Make vs. Buy: Applications Exploiting Sentiment Analysis Services
In reviewing key end-of-year reports, articles and research summaries, a few stand out. Most notably, let’s look at Machine Learning as a Service: Part 1 (Sentiment analysis: 10 applications and 4 services) from Towards Data Science.
Two sections of that June 2018 report stand out:
1. What can I do with sentiment analysis?
This section sets the stage for coming up with potential enterprise use cases. It lists ten good examples from published academic research.
ACTION: IDEATE
- Check the references in footnotes 4 through 13
- Identify all the text streams your enterprise already captures somehow (and other related text streams you might also exploit, such as Twitter commentary that somehow relates to your business)
- Given the text streams, which of the ten use cases might apply to your industry and business?
- Don’t limit yourself to single use cases
- How might you combine use cases to impact your business?
- Consider both opportunity and threat scenarios
- What happens to your business if a competitor emerges that is exploiting these services (or applications that depend on use of such services)?
- Do not make technology implementation assumptions at this stage. (All four vendors listed in this research provide sentiment analysis services from “the cloud.” But there are non-cloud implementations if you need them. Park this issue as you conceptualize potential uses and experiment with the cloud based technology.)
- Don’t limit yourself to single use cases
2. What are some good sentiment analysis services?
This section is a straightforward snapshot of sentiment analysis services. It describes and evaluates Amazon, IBM, Google and Microsoft mid-2018 services. (Details will continue to evolve as the service providers enhance their offerings.)
ACTION: EXPERIMENT WITH THE TECHNOLOGY.
- Assess the feasibility of the business scenarios you’ve envisioned above.
- This is not the same thing as building the “solution” yourselves.
- Â
MOST ENTERPRISES SHOULD NOT BUILD:
- The underlying Natural Language Processing technologies
- Their own sentiment analysis services.
- The production application that relies in whole or part on sentiment analysis services.
Most Enterprises Should:
- Experiment to get a grasp of the feasibility of achieving their business objectives (and better evaluate proposals from suppliers)
- Seek out specialist firms with demonstrated subject matter expertise in their industry
- Follow the section on Five Easy Criteria To Seek described in our blog post on 5 Easy Criteria To Get Quick Returns on AI Investments
Disclosure: I am the author of this article. The opinions are my own. This is not a sponsored post. I have no vested interest in any of the entities mentioned in this post. Neither does the Analyst Syndicate.
Tom Austin
Pretty direct. What’s your take on Part 2 of the Machine Learning As A Service posting?