There are API aggregation sites like Mashape that are already working toward this goal. My startup thing is about using learned models for text processing [1]. Quick tests are also available here [2].
If we are to cite articles here, let me get one from Quim Monzó (a reputed catalan journalist) [1]. I'm pasting a Google translation directly here:
""
Noone can stop it
The latest rumors report that, after the first days of confusion in which many people (myself for example) had just did not understand the superblock Poblenou is so successful that the council considers mounted another even fatter. Instead of the nine islands now recovered and released to the public, the new superblock, which would be the Eixample district, in aggregate compensar- eighties. Sepúlveda go to Paris and Rocafort Aribau, including the four islands of the Clinic. The proliferation of space-and the undeniable fact that trim the circulation paths as important as Roma Avenue, the street Aragon and Great fea- would be a traffic calming infinitely superior to that now allows Poblenou.
In addition, it could hold more activities, not just those that are in the Poblenou. Convert planters of trees of the streets in small urban gardens so residents are planted to tomatoes from parsley. Every afternoon at five, would mime workshops all corners of the street Villarroel. At the intersection of Valencia Viladomat, barbecues because residents can cook sausages grilled tofu and mushrooms in which they exist. Years rain soon, like this, if the mushrooms are imported (Soria, Romania ...) it is verified Mushroom Fair Trade. Without this accreditation we are not allowed to use the picnic tables that are installed because citizens can enjoy these products along with their children. Saying that both barbecues and picnic tables would be designed for students and students of architecture, in a gesture of solidarity and protest sustainable. Once a week, the space freed Gran Via Casanova push forward the Factory of Ideas: a neighbor or neighbor tells a thought that has had this thought and then under discussion between all. It also may explain dreams, which are analyzed equally between everyone.
I can not conclude without noting that bus routes to be amended to avoid going through this supersuperilla would not be fixed. Each driver is free to choose what you want, you also can change from one day to another. This will not only encourage the creativity of colleagues and drivers but the genius of passengers and passenger. That could be targeted opening in September 2018, coinciding with the Feast of Mercy. And I forgot: stretched between the buildings around this area and recovered released settled zip, overflying stops caramelized apples that would allow it (the zip, no stops) than boys and children and adults, and adults could interact with each other and vice versa.
""
Very true, but in the real world where what's urgent goes before what's important, you are often forced to reschedule stuff, and the pen&paper approach clearly is in disadvantage here.
My solution is a dynamic todo called myTinyTodo [1], a very simple app where I can sort stuff by project, deadline, importance, etc.
Ask HN: What value can makers add to ham? I feel I belong to this "new makers" wave and I'm thinking about using Artificial Intelligence to model noise (and cancel it) so as to use cheap SDR dongles more effectively.
I would say that hill climbing with random restarts works better because you must first define a cost function in a precise (and concise) symbolic mathematical way. Very much similar to simulated annealing, stochastic gradient descent and the like. However, GA's can be applied regardless of the piece of knowledge that is required to build this cost function. GA is a universal optimisation technique: as long as you can encode features in a gene form (this is very convenient for binary attributes) and rate the fitting of that gene you are ready to search the optimum.
Key to AI is clear. But, which is the challenge? Are we talking about predicting actually? Or is it rather about building a huge case-based reasoning system that can "replicate" known patterns?
Most recent research is focused on neural nets and deep-learning to that end, this answers the question wrt prediction, but this seems to be a partial solution.
Do you have the gold standard labels of your dataset? Can you ensure that the amount of pos/neg labels is symmetrical?
You can heuristically tune the weights of your lexicon to fit your intuition, but evidence is necessary to progress adequately.
In case you find unbalanced amount of examples, apply an unbalanced effectiveness score like the F-measure to obtain a fair performance of your system.
[1] https://market.mashape.com/atrilla/nlptools [2] http://nlptools.atrilla.net/web/api.php