search

QuantEcon / meta

For issues and discussion covering more than one repository
0 stars 0 forks source link

[paper] Interesting application of Machine Learning #70

Open mmcky opened 1 year ago

mmcky commented 1 year ago

Here is a really interesting pre-print paper that makes use of machine learning

https://osf.io/preprints/socarxiv/uyxh9

HumphreyYang commented 1 year ago

Hi @mmcky,

Thanks for sharing! It is a really interesting paper!

I feel like it is an interesting field where many NLP models can be used. There are many things we could have done, for example, their dataset is drawn from the summarized database where the contextual information is filtered out. If they leveraged contextual information, techniques such as TF-IDF will not be sufficient if they have this type of contextual information have longer corpora, and we can use models such as BERT to subtract more information and conduct deeper qualitative and quantitative analysis.

I think this can also help to improve the time-series analysis where they could have incorporated industry-specific and spatial attributes in the model and analyse industry-specific, time-series, and regional-specific data given there are often policy cycles of easing and tightening in a specific industry in China (like in the semiconductor industry), and different focuses for IP in different regions.