s434730/RandomSec
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

more polish on the clustering dialog

Browse Source 
git-svn-id: http://google-refine.googlecode.com/svn/trunk@264 7d457c2a-affb-35e4-300a-418c747d4874
This commit is contained in:
Stefano Mazzocchi 2010-03-10 01:18:41 +00:00
parent 1000d63539
commit 72b012971f
4 changed files with 38 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/main/java/com/metaweb/gridworks/clustering/binning/BinningClusterer.java
View File

@ -14,6 +14,7 @@ import org.json.JSONException;
import org.json.JSONObject;
import org.json.JSONWriter;
import com.metaweb.gridworks.Gridworks;
import com.metaweb.gridworks.browsing.Engine;
import com.metaweb.gridworks.browsing.FilteredRows;
import com.metaweb.gridworks.browsing.RowVisitor;
@ -52,6 +53,7 @@ public class BinningClusterer extends Clusterer {
if (k instanceof NGramFingerprintKeyer) {
try {
int size = _config.getJSONObject("params").getInt("ngram-size");
Gridworks.log("Using ngram size: " + size);
_params = new Object[1];
_params[0] = size;
} catch (JSONException e) {

8
src/main/java/com/metaweb/gridworks/clustering/knn/kNNClusterer.java
View File

@ -107,7 +107,9 @@ public class kNNClusterer extends Clusterer {
try {
JSONObject params = o.getJSONObject("params");
_radius = params.getDouble("radius");
Gridworks.warn("Use radius: " + _radius);
_blockingNgramSize = params.getInt("blocking-ngram-size");
Gridworks.warn("Use blocking ngram size: " + _blockingNgramSize);
} catch (JSONException e) {
Gridworks.warn("No parameters found, using defaults");
}
@ -143,10 +145,13 @@ public class kNNClusterer extends Clusterer {
}
}
int block_count = 0;
Map<Serializable,Set<Serializable>> clusters = new HashMap<Serializable,Set<Serializable>>();
for (List<String> list : blocks.values()) {
if (list.size() < 2) continue;
block_count++;
for (String a : list) {
for (String b : list) {
if (a == b) continue;
@ -168,7 +173,8 @@ public class kNNClusterer extends Clusterer {
}
}
Gridworks.log("Calculated " + _distance.getCount() + " distances");
Gridworks.log("Calculated " + _distance.getCount() + " distances in " + block_count + " blocks.");
_distance.resetCounter();
return clusters;
}
}

4
src/main/java/edu/mit/simile/vicino/distances/Distance.java
View File

@ -8,6 +8,10 @@ public abstract class Distance {
return counter;
}
public void resetCounter() {
counter = 0;
}
public abstract double d(String x, String y);
}

59
src/main/webapp/scripts/dialogs/facet-based-edit-dialog.js
View File

@ -29,7 +29,7 @@ FacetBasedEditDialog.prototype._createDialog = function() {
var html = $(
'<div>' +
'<div class="facet-based-edit-dialog-controls"><table width="100%">' +
'<div class=""><table width="100%">' +
'<tr>' +
'<td>' +
'Method: <select bind="methodSelector">' +
@ -58,11 +58,11 @@ FacetBasedEditDialog.prototype._createDialog = function() {
'</td>' +
'<td>' +
'<div id="ngram-fingerprint-params" class="function-params hidden">' +
'Ngram Size: <input type="text" value="2" bind="ngramSize" size="3">' +
'Ngram Size: <input type="text" value="2" bind="ngramSize" name="ngram-size" size="3" class="param" datatype="int">' +
'</div>' +
'<div class="knn-controls hidden">' +
'<span style="margin-right: 1em">Radius: <input type="text" value="1.0" bind="radius" size="3"></span>' +
'<span>Block Chars: <input type="text" value="6" bind="ngramBlock" size="3"></span>' +
'<span style="margin-right: 1em">Radius: <input type="text" value="1.0" bind="radius" name="radius" size="3" class="param" datatype="float"></span>' +
'<span>Block Chars: <input type="text" value="6" bind="ngramBlock" name="blocking-ngram-size" size="3" class="param" datatype="int"></span>' +
'</div>' +
'</td>' +
'<td bind="resultSummary" align="right">' +
@ -99,42 +99,33 @@ FacetBasedEditDialog.prototype._createDialog = function() {
self._function = $(this).find("option:selected").text();
$(".function-params").hide();
$("#" + self._function + "-params").show();
self._cluster();
params_changer();
};
this._elmts.keyingFunctionSelector.change(changer);
this._elmts.distanceFunctionSelector.change(changer);
this._elmts.ngramSize.change(function() {
try {
self._params = { "ngram-size" : parseInt($(this).val()) };
self._cluster();
} catch (e) {
alert("ngram size must be a number");
}
});
this._elmts.radius.change(function() {
try {
self._params = { "radius" : parseFloat($(this).val()) };
self._cluster();
} catch (e) {
alert("radius must be a number");
}
});
this._elmts.ngramBlock.change(function() {
try {
self._params = { "blocking-ngram-size" : parseInt($(this).val()) };
self._cluster();
} catch (e) {
alert("radius must be a number");
}
});
var params_changer = function() {
self._params = {};
$(".dialog-body input.param:visible").each(function() {
var e = $(this);
var name = e.attr('name');
var datatype = e.attr('datatype') || 'string';
var value = e.val();
if (datatype == 'int') {
value = parseInt(value);
} else if (datatype == 'float') {
value = parseFloat(value);
}
self._params[name] = value;
});
self._cluster();
};
this._elmts.ngramSize.change(params_changer);
this._elmts.radius.change(params_changer);
this._elmts.ngramBlock.change(params_changer);
//this._elmts.clusterButton.click(function() { self._cluster(); });
//this._elmts.unclusterButton.click(function() { self._uncluster(); });
var left_footer = footer.find(".left");
$('<button></button>').text("Select All").click(function() { self._selectAll(); }).appendTo(left_footer);
$('<button></button>').text("Deselect All").click(function() { self._deselectAll(); }).appendTo(left_footer);