Карта сайта
Версия для печати

На что стоит обратить внимание при выборе инструмента для анализа текстовых данных

30 ноября 2012 На сегодняшний день на рынке представлено немало инструментов для анализа текстовых данных, которые помогают компаниям получать вводные для принятия решений из различных источников, содержащих текст – это могут быть как обычные документы и e-mail переписка, так и посты в социальных сетях. Эксперт компании Gartner рекомендует быть очень осторожным при выборе поставщика подобного решения, так как на качество полученной информации влияет ряд факторов. Ханс Келер-Круер в своем блоге рекомендует внимательно протестировать рассматриваемое решение. Читайте на какие аспекты стоит обращать особое внимание. (Материал опубликован на английском языке)
Giving vendors the third degree is especially important when shopping for text analytics technology because there is a wide range of offerings on the market at various levels of maturity, said Hanns Koehler-Kruener, a research director with Stamford, Conn.-based IT research firm Gartner Inc.

Additionally, text analytics technology is still emerging into the mainstream and therefore terminology and performance expectations will vary from vendor to vendor. As a result, the only real way to find out if a particular text analytics product meets specific needs is through questioning and trial and error, the analyst said.

"It is very much a try and see [situation] -- see if it meets your specific requirements," Koehler-Kruener said. "Don't assume that just because all vendors call something "NLP" (natural language processing) that they all mean the same thing and therefore [results] will be the same."

Text analytics software helps organizations derive potentially valuable business insights from text-based content ranging from word documents and email to postings on social media streams like Facebook, Twitter and LinkedIn.

The technology, which has grown increasingly popular with the rise of social media, typically employs a variety of different mechanisms such as NLP and statistical analysis to deliver results. But that is where the similarities between text analytics products essentially end, Koehler-Kruener said, adding that vendors tend to show the results of text analysis operations in different ways.

While some vendors focus on single use cases, such as social media analysis, e-discovery or voice-of-customer analysis, others offer a range of tools. To add to the confusion, some smaller vendors do not offer "out-of-the-box" software that is ready to use. Instead, they give customers the tools to build their own applications.

In addition to asking questions, Koehler-Kruener encouraged companies to test several text analytics offerings and compare results.

"Test it with your information, not a demo set from the vendor," Koehler-Kruener said.
"We all know that will work perfectly."

The need for new skills

Koehler-Kruener, the co-author of a new Gartner report titled "Who's Who in Text Analytics" that looks at several vendors including SAP, writes that further confusion has resulted from the fact that proper text analysis requires new skills, which can be hard to find.

"Text analytics requires new types of skills to leverage the more advanced and most valuable features of these products, such as combining insight from social media content with customer orders to optimize campaign results and capture new opportunities as they emerge," Koehler-Kruener writes in the report.

Strengths, weaknesses of SAP text analysis software

SAP currently offers two different text analysis products. They are SAP BusinessObjects Data Services, which uses NLP to extract, categorize and summarize information from free-form text sources such as email, documents and notes, as well as data feeds from social platforms.

SAP also has a slimmed down version of the Data Services software through its Rapid Deployment Solution (RDS) program. Released last month, SAP says the software is aimed at helping organizations better understand their customers through sentiment analysis and other functions. Like its more fully functional forebear, RDS software can be used to analyze text on any Web channel with a publicly available API.

One of the key strengths of the SAP Data Services software is its ability to group like terms and concepts among reams of text, Koehler-Kruener said. He also said that SAP supports 31 languages, as opposed to many vendors that only support a handful.

On the downside, Koehler-Kruener said that SAP -- like other vendors -- still hasn't figured out how to truly integrate the results of text analytics operations with more traditional, structured data so that the business has a more "holistic" view of what's going on.

"That’s something where more work is needed," Koehler-Kruener said.

Source: searchsap.techtarget.com