PMI Score Vector

A PMI Score Vector is a real-number score vector that represents the PMI score of each vocabulary member [math]\displaystyle{ t_i }[/math] for some given vocabulary member [math]\displaystyle{ t }[/math] (of a multiset)

• Context:
  • It can be represented a [math]\displaystyle{ [\text{PMI}(t,t_1,\mathbf{C}), \text{PMI}(t,t_2,\mathbf{C}), ..., \text{PMI}(t,t_n,\mathbf{C})] }[/math], where ...
  • It can be used by a PMI-based Distance Function/PMI Vector Distance Function.
  • It can be member of a PMI Matrix.
• Example(s):
  • [math]\displaystyle{ \text{PMI}_{\text{vector}}(``\text{chrome-plated}'', \text{20 Newsgroups}) = [0, 0.6, 0, ..., 0, 0.12] }[/math].
  • …
• Counter-Example(s):
  • a Normed Frequency Vector.
  • an IDF Score Vector.
  • a tf-idf Score Vector.
• See: Text Clustering.

---

---

---