<p style="text-align:justify;">
The effctiveness and scalability of MapReduce-based implementations of complex data-intensive tasks depend on an even redistribution of data between map and reduce tasks. In the presence of skewed data, sophisticated redistribution approaches thus become necessary to achieve load balancing among all reduce tasks to be executed in parallel. For the complex problem of entity resolution with blocking, we propose BlockSplit, a load balancing approach that supports blocking techniques to reduce the search space of entity resolution. The evaluation on a real cloud infrastructure shows the value and effctiveness of the proposed approach.
</p>

<a href="/file/CIKM_LB_poster.pdf" style="float:left; margin-left:20px;">
<img title="Poster@CIKM 2011" src="file/CIKM_LB_poster.png" width="200" height="289" alt="Poster" style="border:1px solid grey;"/>
</a>
<br style="clear:left;"/>

<h2>Keywords</h2>
<ul>
<li>MapReduce, Hadoop</li>
<li>Entity Resolution, Object matching, Similarity Join, Pair-wise comparison</li>
<li>Clustering, Blocking</li>
<li>Data Skew, Load Balancing</li>
</ul>

<h2 id="bibtex_heading">BibTex</h2>
<pre id="bibtex_listing">
@inproceedings{DBLP:conf/cikm/KolbTR11,
author = {Lars Kolb and
Andreas Thor and
Erhard Rahm},
title = {{Block-based Load Balancing for Entity Resolution with MapReduce}},
booktitle = {CIKM},
year = {2011},
pages = {2397-2400},
ee = {http://doi.acm.org/10.1145/2063576.2063976},
crossref = {DBLP:conf/cikm/2011},
bibsource = {DBLP, http://dblp.uni-trier.de}
}
</pre>