In the course of research into the sentiment analysis sector, I learned about Semantria. Founded in 2011, the company is a joint venture between Lexalytics and the software development company Postindustria, as well as the consulting firm DemandGen.
I found the demonstration of the Semantria semantic analysis interesting. You can experiment with the online content processing at https://semantria.com/demo. You can copy and paste a chunk of text as I did. I processed text from my blog Beyond Search and the system displayed a report that is similar in format to Figure 1 on page 18. The idea is that Semantria turns unstructured content into actionable data. The company's website asserts:
"Semantria's API helps organizations to extract meaning from large amounts of unstructured text. The value of the content can only be accessed if you see the trends and sentiments that are hidden within. Add sophisticated text analytics and sentiment analysis to your application. Turn your unstructured content into actionable data."
Semantria's technology integrates with an organization's existing enterprise applications. Those systems can be traditional search engines, enterprise resource planning solutions or content management systems.
I also installed the Semantria Excel add-in. After receiving a "key" to Semantria's API, the semantic functions were accessible within an Excel 2010 spreadsheet. I copied and pasted text into the Excel spreadsheet and explored the different report formats. The information displayed in standard outputs is useful to a person analyzing content from the point of view of a business analyst or a subject matter expert working to figure out nuances in text.
Examples of the utility of the Semantria Excel-based approach are in figuring out what themes or hot buttons reside in open-ended survey results, a collection of customer support Help Desk reports, or gaining insight into a collection of documents written by a particular person. The Semantria system offers both desktop functionality and the company's cloud service.
I did not have to figure out how to access or program the company's API. I just used the Semantria commands added to Excel to process the text. The interface implemented within Excel was easy to use. However, obtaining results requires several clicks for loading, processing and getting them back. That eliminates writing code and having to dig through the documentation for the API.
Performance was under 10 seconds on the two test sets we performed. To be fair, we did not process gigabytes of text. The test corpus was equivalent to several hundred help desk documents. The free API pipes content processing through a throttled service. We used the trial account.
The Semantria content processing system allows a manager "to know what is happening in your content stream." The system extracts entities, themes, sentiment, categories, summaries, facets and relationships. The firm's approach is a semantic thesaurus based on Wikipedia's content. Semantria's service is fully customizable. It uses industry standard software methods, including Dot Net, Python and php, among others.
The company charges on a pay-as-you-go basis. There are no recurring fees or contracts, which is a significant departure from the approach of some other next-generation content processing vendors.
At the CeBIT 2013 show in Hannover, Germany, I was surprised at the strong interest shown in semantics and natural language processing (NLP). In the SAP exhibit hall, senior executives expressed interest in companies offering software that moved beyond key word indexing.
Beyond traditional search
One SAP professional pointed out that traditional search was useful, but there were situations in which the content in the SAP system had to be understood. "We have customers with content in multiple languages, from systems inside the company and from social media, and from Web forms. Search systems are not enough today," the SAP employee said.
The same message was expressed by an IBM (ibm.com) professional. Three Power7 servers glowed in the IBM exhibit hall. I asked about Watson, and I learned that "Watson is a new cognitive system." The German IBM representative had memorized the official lingo on the IBM website:
"Systems like Watson may transform how organizations think, act and operate in the future. Learning through interactions, they deliver evidence-based responses driving better outcomes (see http://www-03.ibm.com/innovation/us/watson)."