Generating paper title joke

Grigori Fursin edited this page Jun 5, 2018 · 5 revisions

[ Home ]

Unfortunately, academic promotion in computer engineering is still mainly based on a total number of publications and citations rather than novelty or practicality of the techniques, availability of code and data, statistical meaningfulness and reproducibility of experimental results.

This means that researchers willing to carefully validate their results and prepare their artifacts for sharing (code, data, models, results, interactive papers) will have to spend considerable extra time and has a disadvantage over those who just want to quickly publish as many incremental papers as possible.

Furthermore, it often results in a bizarre research model where researchers are pushed to spend most of the time on writing their papers (to be published in a top conference or journal) and then quickly fit some experiments no matter whether they are reproducible, practical and support presented idea or not. This, in turn, results in numerous incremental papers that are very well written, published in top conferences and well cited, but based on possibly wrong assumptions and hence without any shared artifacts to hide it. Some typical examples include machine learning based autotuning and run-time adaptation in computer engineering.

Such issues motivated us to create Collective Knowledge Framework where researchers can implement their experiments from shared components and validate them across many shared computational resources (realistic software and hardware). Such artifacts and workflows are shared along with publications to let community validate and extend techniques. Hence, we shift focus from numerous incremental articles to shared artifacts similar to open source software movements.

We hope that CK with shared components will help speed up prototyping of ideas while taking advantage of CK modules for statistical analysis, predictive analytics and experiment validation across many platforms with the help of the community, while promoting publishing of truly novel techniques with shared artifacts (see our vision paper).

However, if you still need to publish numerous papers, we can help you select many exciting topics via our automatic and customizable title generator.

You can also run it from command line as follows:

$ ck pull repo:ck-dissemination
$ ck generate dissemination.publication:template-joke

You can even customize it for your own needs by editing meta description of the dissemination.publication:template-joke entry:

$ ck find dissemination.publication:template-joke
 edit [found_path]/.cm/meta.json file

or

$ ck edit dissemination.publication:template-joke

We also created a related Android-based application that demonstrates how to use our Java-based OpenME interface to connect to CK JSON-based web service to generate and return such title to your mobile device. All source for Android studio are available here.

In the future, we also plan to provide some research paper templates to be able to automatically generate graphs and tables from data continuously obtained during CK software/hardware crowd-tuning.

We also plan to use predictive analytics to automatically detect representative papers out of numerous incremental ones similar to how we find representative data sets and optimizations for programs ;) ...

Have fun!

Questions and comments

You are welcome to get in touch with the CK community if you have questions or comments!

Clone this wiki locally
You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.