Data Integrations

Help you integrate and match various database and data source

Example!

⁠

Search engines have made it easier to conduct literature reviews. However, for some niche topics, search results are often poor.

Snowballing and recommendations from paper repositories may help, but have limitations.

A general tool to search for similar articles without relying on a specific database would be very helpful.

Before that, optimal method to match similar articles needs to be determined.

Four deep learning-based methods were evaluated:

Smooth Inverse Frequency (SIF)

Recurrent Neural Network (RNN)

Attention

Hybrid

They learn a representation to be fed to the actual classifier module, which is a two-layer HighwayNet.

The attention method using only the article title as a feature was found to have the best result.

Execution time:

SIF is significantly faster than the other, while Hybrid is significantly slower.

The Attention method places third but only slightly slower than RNN.

Feature selection:

The title field was found to be the best feature for predicting similar pairs.

The authors’ feature was bad on its own but can improve results when combined with the title.

The keywords feature was found to be almost as good as the title, but combining them did not give significant improvement.

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.