Chapter3: Chemical Topic Modeling - An Unsupervised Approach Originating from Text-mining

reading notes of《Artificial Intelligence in Drug Discovery》

---

1.Introduction

• In chemistry, we tend to describe large sets of molecules using the types of compounds it contains, e.g. peptides. Being able to automatically organize sets of molecules in a similar way would allow chemists to more naturally explore and retrieve information and knowledge from those sets.

2.Topic Modeling and LDA

• There exist several mathematical and algorithmic frameworks for constructing such topic models. They usually share the common idea that the data was generated in a context and that information about the context is latently present in the data. When provided with enough data the most common contexts can be reconstructed.
• Latent Dirichlet Allocation

剩下的内容以后再看