Databases
StackOverflow2013
Postsdbo
POChoose or generate canonical variant from multiple sentences
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POChoose or generate canonical variant from multiple sentences
    primarykey
    data
    text
    <p>I'm working with an API that maps my GTIN/EAN queries to product data.</p> <p>Since the data returned originates from merchant product feeds, the following is almost universally the case:</p> <ul> <li>Multiple results per GTIN</li> <li>Products' titles are pretty much unstructured</li> <li>Products' titles are "polluted" with <ul> <li>SEO-related stuff,</li> <li>information about the quantity contained,</li> <li>"buy two, get one free" offers,</li> <li>etc.</li> </ul></li> </ul> <p><strong>I'm looking for a programmatic way to either</strong></p> <ul> <li><em><strong>choose</em> the "cleanest"/most canonical version available</strong></li> <li><strong>or <em>generate</em> a new one that represents the "lowest common denominator".</strong></li> </ul> <p>Consider the following example results for a single EAN query:</p> <ul> <li>Nivea Deo Roll-On Dry Impact for Men</li> <li>NIVEA DEO Roll on Dry/blau</li> <li>Nivea Deo Roll-On Dry Impact for Men, 50 ml, 3er Pack (3 x 50 ml)</li> <li>Nivea Deo Roll on Dry/blau 50 ml</li> <li>Nivea Deoroller 50ml dry for Men blau Mindestabnahme: 6 Stück (1 VE)</li> <li>NIVEA Deoroller, Dry Impact for Men</li> <li>NIVEA DEO Roll on Dry/blau_50 ml</li> </ul> <p><strong>My homebrew approach looks like this:</strong></p> <ul> <li>Basic cleanup: <ul> <li>Lowercase the titles,</li> <li>strip excessive whitespace,</li> <li>throw out apparent stopwords such as "buy" and "click"</li> </ul></li> <li>Build an array for <code>word =&gt; global occurence</code> <ul> <li><code>"Nivea" =&gt; 7</code></li> <li><code>"Deo" =&gt; 5</code></li> <li><code>"Deoroller" =&gt; 2</code></li> <li><code>…</code></li> <li><code>"VE" =&gt; 1</code></li> </ul></li> <li>Calculate the "cumulative word value" for each of the titles <ul> <li><code>"Nivea Deo" =&gt; 12</code></li> <li><code>"Nivea Deoroller VE" =&gt; 10</code></li> </ul></li> <li>Divide the cumulative value by the length of the title, resulting in a score <ul> <li><code>"Nivea Deo" =&gt; 6</code></li> <li><code>"Nivea Deoroller VE" =&gt; 3.34</code></li> </ul></li> </ul> <p>Obviously, my approach is pretty basic, error-prone and biased towards short sentences with frequently used words – yielding more or less satisfactory results.</p> <ul> <li><strong>Would you choose a different approach?</strong></li> <li><strong>Is there some NLP magic way to take care of the problem that I don't know of?</strong></li> </ul>
    singulars
    1. PO
      singulars
      1. PTAnswer
    1. This table or related slice is empty.
    1. PTQuestion
    1. USvzwick
    1. USvzwick
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PO
      singulars
      1. PTAnswer
    2. PO
      singulars
      1. PTAnswer
    3. PO
      singulars
      1. PTAnswer
    1. VO
      singulars
      1. POChoose or generate canonical variant from multiple sentences
      1. USKenston Choi
      1. VTFavorite
    2. VO
      singulars
      1. POChoose or generate canonical variant from multiple sentences
      1. USvzwick
      1. VTBountyStart
    3. VO
      singulars
      1. POChoose or generate canonical variant from multiple sentences
      1. This table or related slice is empty.
      1. VTUpMod
    1. COWhy wouldn't the words "buy" and "click" show up as statistically common entries in your canonical form?
      singulars
      1. POChoose or generate canonical variant from multiple sentences
      1. USIra Baxter
    2. CO@IraBaxter Updated the question ;)
      singulars
      1. POChoose or generate canonical variant from multiple sentences
      1. USvzwick
    3. CO+1 because you're going to need all the reputation you can get.
      singulars
      1. POChoose or generate canonical variant from multiple sentences
      1. USHassan
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload