I am a linguist building a startup with an English reference software/tool based on Google. It takes a user input, Googles it, processes the results to extract clean examples, and displays them in a way suitable for use in English check and correction. Through many years’ research and trial I have developed the design and result processing algorithms. I am now looking for a developer who can code them into an MVP (minimum viable product).
As well as excellent coding skills, the developer must have expertise in Google search API, Web text processing (especially the ability to identify different elements of search results and Web texts), and natural language processing (especially parse tree navigation). We will start with small modules to build trust. If successful the work can grow into a long-term partnership and even equity sharing if desired.
Below I present a brief pitch for the tool in order to attract talents and convince them that this is a project they can be proud of. I will pay top dollars for excellence, and In return I will screen very rigorously. So please apply only if you are seriously motivated and one of the best with all above qualifications.
I have a limited knowledge of Python, so I would like the tool to be written in Python as much as possible (especially the result processing algorithm) so I can review and tweak the codes if necessary.
The tool helps non-native speakers of English use correct expressions while they write in English. It builds on the current use of Google as the most popular and trusted English reference tool, making it much more convenient and effective for this purpose.
My 2015 user survey with 72 full responses showed the following:
• On average the survey participants spend 21.4% of their English writing time and 93.2 hours a year on checking and improving their English expressions.
• Among all available reference tools including dictionaries and automatic grammar checkers, the one they use most frequently and find most helpful is Google search.
• But only 29% of them find Google “convenient” and are “satisfied with it” as a reference tool, and only 7% of them always get a satisfactory answer from Google.
• Only 59% of the Google query cases reported in the survey found a correct answer.
Dictionaries have a limited content, so users often fail to find what they are looking for, especially long phrases. Moreover, finding dictionary information is a tedious job because it is organized by words, not by expressions. Google resolves these problems with its vast language content where almost any expression is available immediately on demand.
Automatic grammar checkers, as exemplified by MS Word grammar check, are convenient but highly unreliable and have made little progress over the last thirty years of development. For instance, Write and Improve, developed by Cambridge University, currently has an accuracy rate of 10~30%. Automatic checkers will never provide a full solution to language correction because very often it requires the understanding of the context and writer's intended meaning, which are unavailable to computers.
This is why English learners find Google most useful. At the same time, however, they have many grievances with Google because it is not optimized as a language tool. Therefore, optimizing Google for language reference will create the most powerful and most loved reference tool, surpassing Google itself.
My optimization tool enables users to enter and shape their queries with minimum efforts based on what they know about their target expression. It then executes Google searches of appropriate types, extracts clean and relevant example sentences and frequency reports, and displays all search results in a single view with options to add or delete examples and see their contexts. This will not only significantly reduce users’ language check time but more importantly enable them to get results that are not possible with the current manual use of Google.