Patents

Pairwise co-occurence of terms within a 7 word window in the US utility patents on a per-year basis. The date associated with each patent is the year of the priority date of the patent.

The three modes of the tensor correspond to year-term-term, and the values are $log(1+f_{i,j})$, where $f_{i,j}$ is the frequency in which the words $i$ and $j$ appeared in the window. Each $(k,*,*)$ slice of the tensor is symmetric.

Tensor Statistics

Downloadable Files

Citation

None.

License

Discussion