info.xml

<?xml version="1.0" encoding="UTF-8"?>

<info>
<author>Md Iqbal Hossain
  <email>iqbaleric@gmail.com</email>
</author>
<url>https://github.com/LearnSphere/WorkflowComponents/tree/dev/AnalysisLsa</url>
<date>2016-08-14</date>
<abstract>The <b>Domain specific semantic processing portal </b> (DSSPP) performs semantic comparision of natural language inputs (i.e. Student input and ideal answer) </abstract>
<description>
    The LSA component takes in a transaction matrix and outputs LSA similarity values computed from comparing columns specified by header1 and header2. 
    This component is build based on LSA principle. The original domain-specific corpus (TASA) is preprocessed and semantic space (lower dimensional matrix) generated based on that corpus. 
    This component load that pre-generated space to find out similarity between input vectors (columns). This component also support many to many comparison while a bunch of sentences are given as input and returnvals defined as uniqmat. 
    For similarity measure, only cosine distance is available now. Other distance measures will be added next. This component generates a “Output.txt” file which contains the two selected column from Transection matrix and a corresponding LSA score. 
    For More information about <b> DSSPP <b/> please click <a href="http://about.dsspp.com/home/about"><storng>HERE</storng></a>. 
</description>

<inputs>
<b>Transaction export</b>
</inputs>

<outputs>
<b>Text similarity values of selected columns</b>
</outputs>

<options>
<b>col</b> - Option col (returnvals) format returns a column of LSA scores corresponding to the string values in header1 and header2 respectively (if  header1 and  header2 are both specified).<br/>
<b>uniqmat</b> - Option (returnvals) uniqmat format returns an MxN similarity matrix corresponding to the M and N unique string values in header1 and header2 respectively (if  header1 and  header2 are both specified).<br/>
</options>

</info>