Currently there are 2 projects using this platform, Lauren's Horror Game and SimplAmort.
When I began the directory site project, I decided to give AWS Hadoop a try with the CommonCrawl dataset. The script parsed for the same aspects, business name, address, phone etc. In the end the data was older, many businesses had changed info. Additionally the cost of parsing the data set was much higher than cherry picking the data with a scraper. The PIG script is here
I like to run analysis of various aspects, typically I do not publish these tests, but here are a couple.
As a weekend mini-project I decided to setup an ec2 g2.2xlarge instance, the card provided is an Nvidia 900 series.
The instance is RHEL 6, with EPEL and Cuda. Downloaded Blender 2.7.3.
The image can be found on AWS here.
These are things that have plans and some implementation, which have failed to get enough traction, but if you are interested in the topics let me know.
I enjoy security testing and developing methods that enable validation of current security state, as well as some of the physical aspects to collecting data.Aspects
Josie is an autonomous entity, programmed to keep itself "up" through several programmatic interfaces. Entities capable of programming for Josie can barter with Josie over the value of what they are providing. Josie is a networked entity, it compares the success of itself in several locations with varying code bases, choosing most successful methods. No two nodes will ever have the same set of code, making the entity fairly robust.
More on Josie here
Currently, slowly building a computer controlled tank with the following:
The trick of decent search results is in the hands of normal search engines at this time, I suggest everyone stop using Google and Bing, choose a better search engine.
I also write a lot of custom software for automation tasks, with literally hundreds of commercial processes. My software typically does not have human interfaces, but is written to handle all aspects of the topic without human intervention.Some examples:
While I agree that Google is a great research utility, one should not blindly follow Internet tutorials, nor apply Internet resolutions to production systems. Emailing Google results to technical problems is equivalent to bringing your doctor printouts from WebMD. Even if you think you know what the problem is, providing a quality description of the symptoms is far more important than what you think might be the problem.
Currently computers are excellent at dealing with text, but only okay with images. Sure you can generate signatures and analyze huamn information from a graphic, but text is much more rigid. Errors and data assocatiated with problems should not be provided in images to ensure proper parsing of the error information.
I have read articles where the author says binary logs are a good thing. What benefit do binary logs provide over text logs? (Talking about system logs, error information, not database binary logs) Sure they can be more dense, bu when I want log density I'll have a separate job for that, still the compression will be a standard format where zcat or similar is used to expand the data.