(Guest post from IBM Product Marketing Manager, Charith Acharya, IBM SPSS)
If you are like me, and probably 99 % of the population on this planet, chances are that you don’t like filling out a questionnaire. You would find it even more cumbersome if it were open- versus close-ended questionnaire.
An open-ended question involves a lot more effort to answer and is the reason for the added displeasure. But, it’s also the open-ended question, as compared to a close-ended question, which gives a lot more information and insight to a surveyor.
The catch, however, is in painstakingly reading those lines of text and, in a lot of instances, reading in between the lines. To gain insight from those lines of text, we need to find out a way to easily and effectively read those lines, and in turn understand what our customers are actually trying to tell us.
IBM SPSS Text Analysis for Surveys enables organizations to read and analyze those thousands of lines of text generated by their surveys. The product is based on powerful and intuitive Natural Language Process (NL P) – a branch of Artificial Intelligence that analyzes and understands human languages in such a way that it can be coded to elicit useful information. This approach not only offers the speed and cost effectiveness of statistics-based systems but also a far higher degree of accuracy, while requiring far less human intervention.
Let us briefly understand how IBM SPSS Text Analytics for Surveys works.
The first step is to identify and prepare the text (record) to be mined. Preparing the record involves converting the source data into a standardized format. This conversion is performed internally and does not change the original data.
The next step is to mine the record and extract structured data. This involves identifying certain key words or terms, also known as candidate words or terms, which are used to identify concepts in the record. Next, types are assigned to these concepts. A type classifies a concept -- for example, a positive or a negative word, first name, etc. For this classification, compiled resources and built-in libraries are used. However, additional types can also be defined by the user. After a couple more steps, text mining algorithms are used to extract the data.
The concepts and types that were extracted in the previous step are then used to generate categories so that the record can be assigned to a particular category if the record contains text that matches an element of a category's definition. Categories and their definitions are generated through one of the built-in techniques. Assigning a record is done through a process known as scoring whereby unique identifiers are assigned to the category definitions for each record or document.
The structured data is then analyzed using traditional data mining techniques, such as clustering, classification, and predictive modeling, to discover relationships between the concepts.
IBM SPSS Text Analysis for surveys is not limited to traditional open-ended surveys. The text from the numerous social media sites, on-line clubs and the like, similar to the answers of an open-ended question, can divulge a wealth of information. This is possible by simply choosing the concerned web feed as the input instead of a traditional file such as MS Excel.
IBM SPSS Text Analytics for Surveys supports Chinese Simplified, English, French, German, Italian, Japanese and Spanish. In addition, the product also enables you to translate text from a list of supported languages, including Arabic and Persian into English. This feature enables you perform text analysis on the translated text and then deploy these results to people who could not have understood the contents of the source languages.
The avenues for using IBM SPSS Text Analytics for Surveys are endless given the social and increasingly expressive world that we live in. We just need to be aware that intelligent tools are able to understand human text and the underlying emotions of those texts exists!
Click here to download either a FREE 14-day Trial Version or resources for of IBM SPSS Text Analytics for Surveys.