Misspellings and their impact on the analysis of online reputation
We are witnessing a dizzying development of an increasingly global and digital world. The upsurge of the communications has changed the way in which people are in contact, communicate and inform. For this reason, in turn, the tools and technological services are directed to manage the information that the users give out on the Internet in order to store, monitor and measure that information that appears in any way and medium. The final aim is to know the online reputation of a brand, company or person.
Which company would not want to know, for example, what the users say about its key product or which is the impact and the opinion that a person has generated after a meeting?
Certainly, this information gives an incalculable value to a company for managing its marketing strategies.
Along with the need of the companies to know its online reputation, new companies and technological services emerge with the aim of analysing the comments and opinions of the users, monitoring the presence of the companies in the media, etc. to inform the final client about what the users say about them on the media and set a course of action for the future.
However, the task of storing, monitoring and measuring the information that a client has given out on the media becomes difficult if the information to be retrieved is linguistically incorrect. That is, the users (whether a company or a client) make several errors such as shifting the letters, putting words together, or misspelling. This makes the task of processing natural language difficult, since the computer can’t understand what a user or a company wanted to say at some point.
The task of transforming that confusing and partially incorrect text into a text that can be processed by the computer will be the first one, and it is essential.
Which are the consequences of a text that contains misspellings?
Imagine that a company, for example a restaurant, wants to know what their clients say about its dishes and beverages or even about the hygiene. If the misspelled words are not corrected, it is difficult to retrieve all the information about the key words of the comments (for example, ‘wine’, ‘omelette’, ‘hamburger’).
Thus, it can be say that if misspellings are not corrected, part of the information will be missed and, therefore, the reports generated by the online reputation tools will lose objectivity.
Which would be the method of analysis?
The first step, undoubtedly, is to retrieve the data that will be analysed, that is to say, to store the data and monitor the media that will be analysed. The second step is to correct the misspellings of the thousands or millions of comments stored and monitored automatically. Correcting misspellings manually is not an option, since the amount of comments to be examined makes this task costly (on time and costs). This second step gives room to the quantitative and qualitative objective analysis of the data.
To sum up, it has been shown that spelling correction as well as the analysis of the spelling phenomena of the comments, posts, etc. is an essential task. Only in this way, online reputation can be analysed and measured with an excellent quality. This also guarantees that all the information is retrieved from the Internet.