Voice of Customer , Sentiment analysis & Feedback service
Some time ago, I was watching a game on TV and the broadcaster started comparing the two teams by showing what fans, people are talking about these teams and players on social media . Broadcasters excitedly compared the percentages and numbers of tweets of the teams and flashed which marquee player is currently more trending.
It left me wondering with more deeper questions, is this public sentiment really that important for the determination of the play outcome or it has really become important enough to analyse , even if it is not as vital as players and teams skills or their strengths & weakness , their game tactics & strategy ? or is it important as any other game analysis attribute ?
If a Presidential debate is happening or a new movie is launched or for example when a new consumer product is launched , I can see value of such similar analysis on what people are talking about it as an important feedback.
Voice of Customer – Voice of the customer – Wikipedia, the free encyclopedia has a intrinsic value, always have , always will be.
Sentiment analysis is important and can not be ignored how so ever fickle it may sound initially, if probed deeper invariably you may find the value even if you take tweeter churning by TV broadcaster during a game , as it is perhaps important for the sponsors and marketing campaign around the players, teams and sports and perhaps for hard loyal fans and perhaps it wins over loyalty of some bystanders.
Business’ would be always interested to get deeper insight into market trends and customer perception of their brand and products. They would like to Proactively respond to customer sentiment for improved brand loyalty, for stronger customer relationships & use it with their marketing campaign to guide business strategy.
The social media data is unstructured, voluminous and fast changing , is it considered big data , perhaps , I could imagine any solution which would do analysis on this data, have to do in real time text data processing.
Incidentally HANA does have text analysis capabilities in its native platform. HANA developer guide has text analysis section detailing how to use these options using SQL commands. One has option to build applications that can do these text and sentimental analysis using the indexed tables.
Text from Social media, sources like Facebook, Twitter can be imported in HANA Tables including from a Hadoop system using smart data access. The content imported for analysis may not be character string , it could be HTML, XML strings or Word , PDF documents.
Although not truly real time, still Text Analysis Index can be easily built on these HANA table containing social media contents like twitter messages, using native HANA text analysis based fulltext index creation busing Linganalysis technique to derive words into tokens as nouns, verbs, adjectives, propositions, puncuations etc. This way message is split into different segment of data and stored in an index table.Query to find out which noun is appearing most in this table can easily be relayed for example as most twitted or talked or trending player .
Messages , for example “New Yorkers like Riley” convey some sentiment in the message. Words like Great , wow for that matter profanity words also carry sentiments. Words can be broken into different token types like person, city, organization, topic, and sentiment like as in weak positive sentiment or strong positive sentiment and even weak or strong negative sentiment , by considering different tokens including emoticons. Native HANA allows building of Index table on these text messages using voice of customer configuration so more intelligent analysis can be done to for example indicate if social media is cheering for their team or cheering which team more by looking at token values containing sentiment.
Another great native feature available on HANA Cloud Platform is feedback service which allows to collect end user feedback. It provides predefined analytics on the collected feedback data – feedback rating distribution and detailed text analysis of user sentiment (positive, negative, or neutral) as illustrated below.
Web application , example shown below provides option to give feedback.
HTML Form data is passed as AJAX Post request .
All feedback is collected by the HANA Cloud application and it can be further analysed for sentiment analysis.
The applications allows probing the Feedback by Positive, Neutral, Negative and Request or Problem type kind of sentiment.