w2v-server: improve table head readability
diff --git a/templates/index.html.ep b/templates/index.html.ep
index 1e57de8..215ccdf 100644
--- a/templates/index.html.ep
+++ b/templates/index.html.ep
@@ -341,7 +341,11 @@
 			    <tr>
 				    <th align="right">#</th><th align="right">cos</th><th align="left">paradigmatic</th>
             % if($collocators) {
-				      <th title="The window around the target word that is considered for summation. Absolute value can be too low because of sub-sampling frequent words.">w</th><th align="right" title="&#34;Responsivenes&#34; of the collocator at the relative position @. Approximation of the probability that the combination of the target word and the collocator at the relative position @ come from the corpus.">resp.</th><th title="Probability of the collocator at window location @."align="right">p(c<sub><small>@</small></sub>)</th><th align="right">Σp(c<sub><small>@</small></sub>)/|w|</th><th align="left">syntagmatic</th>
+				    <th title="The window around the target word that is considered for summation.">w'</th>
+						<th align="center" title="Raw (max.) activation of the collocator in the output layers.">a</th>
+						<th title="Σp(c<sub><small>@</small></sub>) – Sum of the probability approximations that the combination of the target word and the collocator at the relative position @ come from the training corpus. Single approximations can be distorted because of sub-sampling frequent words and the sum cannot itself be interpreted as probability."align="center">Σp</th>
+						<th align="right">Σp/|w|</th>
+						<th title="c" align="left">collocator</th>
 				    % }
 			    </tr>
 			    % my $j=0; my @words; my @vecs; my @ranks; my @marked;