diff --git a/src/explore.ipynb b/src/explore.ipynb
index 3578be77528fe349a6cf0aee634a6f9cbb66e290..ef63de256bece4089b3a6a22b1d2593050a7a1b5 100644
--- a/src/explore.ipynb
+++ b/src/explore.ipynb
@@ -49,6 +49,22 @@
     "df_origin = pd.read_csv(\"quarry-32518-all-filters-sorted-num-hits.csv\", sep=',')"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "And the dataset with the 2nd round of labeling:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df_2nd = pd.read_csv(\"20190106115600_filters-sorted-by-hits-manual-tags-2nd-round.csv\", sep='\\t')"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -308,7 +324,12 @@
     "# --> so available groups are \"default\" and \"feedback\"\n",
     "# TODO: question: what do they mean?\n",
     "# From https://www.mediawiki.org/wiki/Extension:AbuseFilter/abuse_filter_table :\n",
-    "# \"The group this filter belongs to, as defined in $wgAbuseFilterValidGroups.\" still don't get it"
+    "# \"The group this filter belongs to, as defined in $wgAbuseFilterValidGroups.\" still don't get it\n",
+    "'''\n",
+    "Feedback seems to be a legacy thing from a plugin which was supposed to collect readers feedback on articles.\n",
+    "It wasn't deemed useful enough though and was eventually turned off.\n",
+    "All the filters that referred to it are in the meantime marked as deleted.\n",
+    "'''"
    ]
   },
   {
@@ -490,6 +511,15 @@
     "\n"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "It is noticeable that quite some filters were last modified in June 2016 (93 filters). Potentially, a lot more filters were modified at this time and also modified later again.\n",
+    "An inquiry into the filters history (https://en.wikipedia.org/wiki/Special:AbuseFilter/history?user=&filter=669) shows that an edit filter manager did a big cleanup and (mostly) deleted stale filters.\n",
+    "All of this doesn't seem to be particularly exciting or worth discussing though."
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -499,7 +529,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [
     {
@@ -633,6 +663,7 @@
    ],
    "source": [
     "# filter hits per month (all filters) (data quarry from 05.03.2019, that's why hitcount Mar 2019 is so small)\n",
+    "# Mar 2009 is also lower, since the first filter was introduced on 17.3.2009 and not at the beginning of the month\n",
     "df_hits = pd.read_csv(\"quarry-34014-how-many-abuse-filter-hits-have-been-there-per-month-en-wiki-run346197.csv\", sep=',')\n",
     "df_hits['LogMonth'] = pd.to_datetime(df_hits['LogMonth'], format=\"%Y%m\")\n",
     "\n",