What is the sorting method used in IEEE Xplore ?

[sajiv]

Simply, IEEE Xplore search results are weighted and sorted by relevance based on an algorithm using the following three parameters:

    * Frequency of the search term in the full abstract/citation record
    * Text length of the abstract/citation record
    * For AND queries, proximity of the search terms throughout the abstract/citation record

More on Verity search operators:

All Verity search operators return 1.0 or 0, with the following exceptions: <MANY>, <NEAR>, <NEAR/N>, <AND>, <OR>, and <ACCRUE>.

<MANY> is a modifier for the evidence operators (<WORD>, <STEM>, <PHRASE>, <SENTENCE>, <PARAGRAPH>) that generates a score proportionate to the number of times the evidence appears in the document compared to the size of the document.

<NEAR> and <NEAR/N> generate a score depending on how close together the words appear in the document. The score will be 1.0 if the words are next to each other.

<AND> returns the minimum score of all its children.
<OR> returns the maximum score of all its children.
<ACCRUE> derives the score from the scores of all its children.

Children with a score of 0 do not contribute to the score. Note that children often default to a weight of 0.5, so you will often see lower scores with an <ACCRUE> compared to <AND> and <OR> the relative ranking of the documents is usually more important than the actual score. You will get the best results by keeping weights in the range of 0.2 to 0.8 for the children of <ACCRUE>.