update.

Author: lsvih

demo/README.md

@@ -7,6 +7,14 @@ pip3 install -r requirements.txt
 python3 app.py
 ```
 
+## Installation Preconditions
+
+Before starting-up this demo, please make sure the gStore has been installed and started.
+
+The gStore can be installed by following the instructions in [gStore Installation](https://github.com/pkumod/gStore).
+
+Besides, the RDF-VG dataset should be loaded into database before running this demo (refer to RDF-VG Builder).
+
 ## gStore Connection Config
 
 Please modify the `gStore_config` variable to connect to your gStore database.
@@ -48,7 +56,7 @@ python3 app.py
 
 Example SPARQL
 ```sparql
-SELECT ?img ?x ?y ?h ?w ?dog_obj WHERE { 
+SELECT ?img ?x ?y ?h ?w ?dog_obj WHERE {
 	?dog_obj <hasName> "dog"^^<http://www.w3.org/2001/XMLSchema#string>.
 	?cat_obj <hasName> "cat"^^<http://www.w3.org/2001/XMLSchema#string>.
 	?dog_obj ?mm.sim ?cat_obj.
@@ -64,7 +72,7 @@ SELECT ?img ?x ?y ?h ?w ?dog_obj WHERE {
 
 Converted SPARQL
 ```sparql
-SELECT ?img ?x ?y ?h ?w ?dog_obj WHERE { 
+SELECT ?img ?x ?y ?h ?w ?dog_obj WHERE {
 	?dog_obj <hasName> "dog"^^<http://www.w3.org/2001/XMLSchema#string>.
 	?img <hasObject> ?dog_obj.
 	?dog_obj <x> ?x.

parser/README.md

@@ -0,0 +1,7 @@
+# VGStore parser
+
+This is a simple implementation of a parser for the VGStore.
+
+By this parser you can parse custom predicates and restore the standard SPARQL queries.
+
+

rdf-vg/README.md

@@ -1,3 +1,27 @@
 # RDF-VG Builder
 
-WIP 
+## Introduction
+
+This is a tool for building a RDF-stored Visual Genome dataset.
+
+The data structure of Visual Genome is refer to [Visual Genome Python Driver](https://github.com/ranjaykrishna/visual_genome_python_driver).
+
+The ontology of RDF-VG is described in our paper and this repo.
+
+
+## Usage
+
+1. Download the [Visual Genome](https://visualgenome.org/) dataset and unzip it.
+2. Install requirements: `pip install -r requirements.txt`
+3. Run rdf-vg builder by `python3 process.py`
+4. Deduplicate the triples by `python3 clean_dup.py`
+
+Now you are supposed to get a series of `.nt` files which contain the RDF triples of Visual Genome.
+
+## Import RDF-VG into gStore
+
+1. Install gStore (refer to [gStore](https://github.com/pkumod/gStore)).
+2. run `./bin/gbuild -db vg -f xxxxx.nt` to import RDF-VG nt triples into gStore.
+3. Serve RDF-VG by `./bin/grpc -db vg -p 5001`.
+
+Now you can use the RDF-VG dataset by gStore query interface (SPARQL).