Predictive Texting

To err is human and to typo in predictive texting is smart phones! 🙂

This is my first effort to share my blog on LBC (Loose Bloggers Consortium). This topic was suggested by Padmum, for the weekly Friday Loose Bloggers Consortium where currently nine of us write on the same topic every Friday.  I hope that you enjoyed my contribution to that effort.  The eight other bloggers who write regularly are, in alphabetical order – Ashok, gaelikaa, Lin, Maxi, Padmum, Rummuser,  Shackman and The Old Fossil. Do drop in on their blogs and see what their take is on this week’s topic. Since some of them may post late, or not at all this week, do give some allowance for that too!

I was aware about LBC, thought of joining the group earlier too. However, I write only once a week on my blog.  It is a big responsibility to write and on regular basis is a very tough job with the kind of engagement I have. Honestly it has been difficult to keep up with my regular blogs of Business and the Buddha.

1. First reason is, I have been doing this mistake often. Here is one – Writing “Vegetable” instead of “Venerable” to Ramana uncleji is one. In my Whatsapp, I have written most often – *<WORD> TYPO <CORRECTED WORD>. One can do an analysis on the same and infer that one of the most frequent word from me is “TYPO” or a typo – a misspelling.

2. I have been doing text analytics so this field is of some interest for me. Predictive texting learns from an individual’s usage of words as well. I can save “Pravin”, “Uncleji” as right spellings and the program will learn. In fact, with more knowledge of my usage of words it may predict what is coming. The type of analytics I am doing is on Twitter feeds & text based identification of sentiments as a starter, instead of predicting text.

Coming to the point – How did predictive texting start? Do you remember Nokia mobile phones used to have T9 feature. This was a dictionary, T9 used to do auto suggest. When user used to type a digit (in those days digits used to have 3/4 alphabats associated) T9 used to offer suggestion. The same is extended to predictive texting of current edge.

Our machines have become smart to beat humans in Jeopardy quiz but still Predictive texting is error prone. Is also referred as “Cupertino effect” because in the initial year predictive text used to change cooperation to Cupertino :).

We should not be blaming machines completely for these errors, after all they are not humans and in whatever case they are also learning. Machine learning is a concept where computers learn from past. Recently published news suggest what Google is trying in – Machine Learning. One thing they are doing and continue to do is – suggestions based on the user behavior. Just imagine how interesting it would be to have machines helping us in whatever endeavors

Be patient  something worthwhile is in  the offering. Please Cupertino (cooperate) – as vegetable (venerable) uncleji did.