predict pubmed

README.md

@@ -183,29 +183,23 @@ sh pubmed.sh preprocess [your_data_name]
 sh pubmed.sh config [your_data_name] [model_name] [gpu]
 ```
 
-## 4.4. Predict
-
-1. Preprocess raw text
+2. Predict
 
-- Tokenize raw text and prepare data for prediction
-```bash
-sh preprocess.sh raw
-```
+- [your_data_name] = my-pubmed
 
-2. Predict
 ```bash
-sh run.sh predict [task] raw text
+sh pubmed.sh predict [your_data_name]
 ```
 
 3. Retrieve the original offsets
 
 ```bash
-sh run.sh offset [task] raw text
+sh run.sh offset [your_data_name]
 ```
 
 - Check the output in
 ```bash
-experiments/[task]/predict-raw-text/ev-last/[task]-brat
+experiments/[your_data_name]/results/ev-last/[your_data_name]-brat
 ```
 
 # 5. Visualization
@@ -229,9 +223,9 @@ python standalone.py
 
 2. Copy the predicted data into the brat folder to visualize
 
-- For the raw text prediction
+- For the raw text prediction: [your_data_name] = my-pubmed, [model_name]=cg
 ```bash
-sh run.sh brat [task] raw text
+sh run.sh brat [your_data_name] [model_name]
 ```
 
 - Or for the shared task
@@ -243,8 +237,9 @@ sh run.sh brat [task] gold test
 3. Visualize
 
 - The data to visualize is located in
+
 ```bash
-brat/brat-v1.3_Crunchy_Frog/data/[task]-brat
+brat/brat-v1.3_Crunchy_Frog/data/[your_data_name]-brat
 ```
 
 # 6. Acknowledgements

pubmed.sh

@@ -52,6 +52,49 @@ elif [ "$TASK" = "config" ]; then
 
     python scripts/generate_configs.py $EXP_DIR $MY_DATA $MODEL_NAME $GPU
 
+# predict
+elif [ "$TASK" = "predict" ]; then
+    echo "Predict: "
+
+    MY_DATA=$2
+    EXP_DIR="experiments/$MY_DATA"
+
+    # predict
+    python predict.py --yaml $EXP_DIR/configs/$TASK-$MY_DATA.yaml
+
+# retrieve offset
+elif [ "$TASK" = "offset" ]; then
+
+    echo "Retrieve original offsets: "
+
+    MY_DATA=$2
+
+    # paths
+    REFDIR="data/$MY_DATA/processed-text/$MY_DATA-text" # reference gold data
+    PREDDIR="experiments/$MY_DATA/results/ev-last/ev-tok-a2/"
+    OUTDIR="experiments/$MY_DATA/results/ev-last/" # retrieve the original offsets
+
+    python scripts/postprocess.py --refdir $REFDIR --preddir $PREDDIR --outdir $OUTDIR --corpus_name $MY_DATA --dev_test pubmed
+
+# prepare data for brat
+elif [ "$TASK" = "brat" ]; then
+
+    echo "Prepare data for brat"
+
+    MY_DATA=$2
+    MODEL_NAME=$3
+
+    PRED_DIR="experiments/$MY_DATA/results/ev-last/$MY_DATA-brat/"
+    BRAT_DIR="brat/brat-v1.3_Crunchy_Frog/data/"
+
+    # annotation file
+    CONFIG="configs/brat/$MODEL_NAME"
+    if [ -d $CONFIG ]; then
+        cp $CONFIG/* $PRED_DIR
+    fi
+
+    # brat
+    cp -r $PRED_DIR $BRAT_DIR
 fi