Basic Examples (5) 

The first article ever on ArXiv:

In[1]:=

Out[1]=

In[2]:=

Out[2]=

In[3]:=

Out[3]=

In[4]:=

Out[4]=

A DateListPlot showing the trends in the most popular title words in theoretical physics category ("hep-th", primary or cross-list):

In[5]:=

Out[5]=

All the 50 most common 2-neighbour title words on the whole ArXiv, ever:

In[6]:=

Out[6]=

Authors with more than one possible name (and categories) are conveniently registered as "ArXivAuthor" entities. For example:

In[7]:=

Out[7]=

We can then also easily create an author citations graph, with the tooltip indicating the articles ids:

In[8]:=

Out[8]=

Scope (13) 

The dimensions of the whole ArXiv main dataset (at the end of September 2024):

In[9]:=

Out[9]=

Let us create a super-database with all computer science "cs" type (primary or cross-list) categories:

In[10]:=

and then let us visualize the most and less frequent title words:

In[11]:=

Out[11]=

Let us calculate the 10 most frequent categories, with their meaning and number of articles each:

In[12]:=

Out[12]=

We can create train and test sets using only 5000={4500,500} titles and abstracts for each category:

In[13]:=

we can train a NN to classify these categories, with layers' dimension 80 and dropout level 0.5:

In[14]:=

Out[14]=

In[15]:=

Out[15]=

Even with a basic 30 minutes training on laptop CPU, we obtain 89% accuracy:

In[16]:=

Out[16]=

and a rather clean confusion matrix:

In[17]:=

Out[17]=

We could even classify authors within the same category, with ArXivClassifyAuthorNet.

Extracting TEX introduction:

In[18]:=

Out[18]=

also TEX formulae:

In[19]:=

Out[19]=

Explain a technical concept using an article introduction and LLMSynthesize:

In[20]:=

Out[20]=

Let us visualize all authors with more than 7 papers, in primary category "cs.NA":

In[21]:=

Out[21]=

Let us pick a random author among them and use LLM functionality to explain his overall work:

In[22]:=

Out[22]=