A while ago I wrote a post discussing some of the shortcomings of current approaches to search; focusing on the relatively primitive and inefficient formalisms used by many database vendors to express search strategies. In that post, I argued that the conventional approach shares many of the shortcomings of early programming languages such as first generation Basic, relying on arbitrary labels such as line numbers to convey structure and organisation.
Of course, different database vendors offer their own syntactic variant (and that in itself is a source of error and inefficiency), but I maintain that at a lexical and semantic level, the formalism by which most search strategies is expressed is obfuscated and inefficient. My argument was (and is) that the conventional approach remains popular at best for reasons of inertia, and at worst for reasons of monopoly, since there is as yet no practical alternative. Indeed, if you consider the evolution of programming languages (and software engineering techniques in general) since the days of first generation Basic, then the ossification of this approach becomes even more apparent.
Case in point: can you interpret the semantics of a strategy such as this? Moreover, can you identify the error, or spot the redundancy?
Likewise, could you simplify an expression such as this?
((AI OR ‘machine learning’) AND (NOT ‘neuro-linguistic programming’ AND NLP) OR NLP)
There’s plenty more to say on this topic in my previous posts, but in this post I’d like to focus instead on solutions. So you perhaps won’t be surprised to learn that we’ve been working on an alternative, which we call 2dSearch. We’ve recently launched the website, the app, and a back-end NLP service (more on that later). For now, just point your browser to:
Like any prototype, it’s a work in progress and we welcome feedback. Moreover, it’s an MVP, and in that respect represents a just a fraction of things you might expect to see in a polished product. We have a host of things in the pipeline, and I’ll be sharing details of those in due course.
But in advance of that, we welcome first impressions, both positive and negative, and attempts to ‘break’ it. Feel free to respond directly via the in-app feedback widget, or in the comments box below. We should also acknowledge the support of Innovate UK whose R&D grant funding has been instrumental in bringing this to fruition. There’s a lot more I could say, but that’s probably enough for now. If anything is unclear, have a look at the FAQ, or just ping me a note.
Let me know what you think!
[PS and if you think you know the answers to the brainteasers above, feel free to share them below :)]
Leave a Reply