Oops I did it again!

You don’t need to be able to type a hundred words a minute to make regular spelling mistakes. Whether it’s habitually putting certain letters the wrong way round or adding an extra character at the end of a word, we’ve all suffered from faulty fingers at some point. But how does a virtual assistant recognize these kinds of errors and what happens when we try, incorrectly, to spell unfamiliar or made-up words such as a product name?

With some chatbots or virtual assistants one wrong letter can be the difference between delivering the right answer or disappointing the user. Yet it needn’t be the case. As humans, we can normally recognize the word someone meant to spell. Sometimes it’s an obvious mistake where two letters have been transposed such as ‘teh’, while others are not quite so clear such as ‘juxtaposw’, that might occur when the user hits the key next to the correct one.

However, by accounting for common mistakes such as spelling distances or hitting the same key twice, it is possible to ensure that an intelligent virtual assistant is able to understand even the biggest of fat finger blunders. In addition, by including context of the sentence it is also possible to recognize other common spelling errors such as ‘there’ instead of ‘their’.

The difficultly for a lot of natural language solutions comes when the user has to spell words that are unfamiliar such as a made-up product name or a foreign language and quite possible one that includes other special characters not used by  the native speaker such as the French œ or the Swedish ö. In these types of situations an intelligent virtual assistant is able to convert typical symbols to account for any mistakes or inability to type certain characters.

But sometimes, despite the various ways to account for human error, we persist on challenging the virtual assistant by spelling a word completely incorrectly. In these cases, where words are systematically spelled incorrectly by users, there may be no other way than to teach the virtual assistant what the user is trying to tell them – reinforcing their belief that sometimes humans just aren’t as intelligent as them.

Leave a Reply

Your email address will not be published.

top