Parsing Syntax Accurately: Jeff Caitlin at SAS 2015

  1 m, 48 s

parsing syntaxSentiment Analysis Symposium

Lexalytics is proud to have sponsored  the Sentiment Analysis Symposium in New York earlier this month. SAS featured both speakers and workshops aimed at keeping up with sentiment analysis technology and discovering the business value in opinions, emotions, and attitudes. Speakers included representatives from Verizon, Instagram, Bottlenose, as well as our very own CEO, Jeff Caitlin, who spoke on the importance of accurately parsing syntax.

The Three Components to Meaning

There are three important components to consider when analyzing text to determine its meaning.

-Semantics: The definition of a word depending on syntax and context.

-Context: The circumstances surrounding the text, including the speaker, what they’ve said in the past, and in what particular space they are speaking.

-Syntax: The effect of sentence structure on the meaning of a word or phrase.

The Importance of Syntax

The ability to parse syntax correctly is vital for accurate sentiment analysis. It’s not enough for a program to simply weigh individual words for their sentiment. Sentences can contain several different entities towards which different sentiments are being expressed, depending on how the sentence is ordered.

For example, the following three sentences contain almost identical wording, but their order determines the sentiment attached to both entities – Apple and Steve Jobs:

  • Apple was doing bad until Steve Jobs returned.
  • Because Apple was doing bad, Steve Jobs returned.
  • Apple was doing bad because Steve Jobs returned.

The Solution to Parsing Syntax

To a machine, there are many logical ways for parsing syntax that simply don’t make sense to a human. For example, the sentence “John went to the store for milk”, might be interpreted as John going to the “store for milk” or milk store, as opposed to going to the store with the intent to buy milk. Humans automatically interpret the sentence as meaning the latter, but machines don’t.

Since humans naturally parse syntax, the best way to ensure sentiment analysis software will parse syntax appropriately is to learn like a human. To learn more about parsing syntax accurately and what that entails, check out Jeff Caitlin’s presentation below.