Update README.md

Author: godfanmiao

ST_DM/KDD2022-DuMapper/README.md

@@ -7,9 +7,10 @@ To save the tremendous labor costs, we devised DuMapper, an automatic system for
 
 ### How to run
 
-DME:
+Deep multimodal embedding (DME) of a POI: This neural module takes the signboard image and the coordinates of a POI as input to generate its multimodal vector representation. 
+
+Approximate nearest neighbor (ANN) search through the large-scale POI database: This module takes advantage of ANN algorithms to conduct a more accurate search through billions of archived POI embeddings in the database for verification within milliseconds.
 
-ANN:
 train.py: train and build DME model.
 
 infer.py: generate multimodal embedding of a POI.