diff --git a/mongo/Dockerfile b/mongo/Dockerfile
index 21df1d3..3dfd9ce 100644
--- a/mongo/Dockerfile
+++ b/mongo/Dockerfile
@@ -1,6 +1,7 @@
 FROM python:3.9
 RUN mkdir /app
 COPY requirements.txt /app
-RUN pip install -r requirements.txt
+RUN pip install -r /app/requirements.txt
 COPY *.py /app
 WORKDIR /app
+ENTRYPOINT ["python", "./mongocrawler.py"]
diff --git a/mongo/mongocwarler.py b/mongo/mongocrawler.py
similarity index 96%
rename from mongo/mongocwarler.py
rename to mongo/mongocrawler.py
index db781a0..9eb738d 100644
--- a/mongo/mongocwarler.py
+++ b/mongo/mongocrawler.py
@@ -278,8 +278,7 @@ def get_link_features(link):
     for i,feature in enumerate(features):
         if len(feature) < 1:
             continue
-        if feature.isdigit():
-            feature = "<NUM>"
+        feature = re.sub("[0-9]","*",feature)
         res.append(str(i)+ "-" + feature)
     if len(res) < 2:
         return None
@@ -322,6 +321,9 @@ class LinkClassifier:
     def test(self,testset):
         # eval
         gg = 0
+        true_positive = 0
+        positive = 0
+        false_negative = 0
         for item  in testset:
             l = item["url"]
             cl = 0
@@ -331,12 +333,19 @@ class LinkClassifier:
             r = 0
             if pcp > 0:
                 r = 1
+            if cl == 1:
+               if r == 1:
+                 true_positive += 1
+               positive += 1
+            if r == 1 and cl == 0:
+                    false_negative += 1
             if r == cl:
                 gg += 1
             else:
                 print("MISS",l,cl,pcp)
-        print("Accuracy:")
         print(len(testset))
+        print("Precision: {}, Recall: {}".format(true_positive/positive,true_positive/(true_positive+false_negative)))
+        print("Accuracy:")
         acc = gg / len(testset)
         print(acc)
         return acc
@@ -407,6 +416,7 @@ def link_summary(db,hostname):
     goodcount = 0
     info = {}
     crawled_count = 0
+    bad_crawl_count = 0
     for item in res:
         count = item["count"]
         st = item["_id"]
@@ -415,7 +425,11 @@ def link_summary(db,hostname):
             goodcount += count
         if st != "frontlink" and st != "backlink": 
             crawled_count += count
+            if st != "good":
+                bad_crawl_count += count
         info[st] = count 
+    info["crawled_count"] = crawled_count
+    info["bad_crawl_count"] = bad_crawl_count
     baclink_cout = 0
     if "backlink" in info:
         backlink_count = info["backlink"]
@@ -469,7 +483,7 @@ def sample_links(db,hostname,status,batch_size):
     for item in res:
         for item in res:
             cll = cl.classify(item["url"])
-            cll += random.uniform(-0.1,0.1)
+            #cll += random.uniform(-0.1,0.1)
             sample_links.append((item["url"],cll))
             if cll > 0:
                 predicted_good += 1
@@ -521,6 +535,7 @@ def createdb():
     htmlcol.create_index("html_md5",unique=True)
     domaincol = db["domains"]
     domaincol.create_index("host",unique=True)
+    domaincol.create_index("average_fetch_characters",unique=True)
 
 @cli.command()
 @click.argument("link")