Relation Extraction Verb Actions

AlchemyAPI provides the ability to extract and normalize "verb actions" within identified Subject-Action-Object relations, enabling identification of buying signals, key events, and more.

What are Verb Actions?

Verb actions are normalized versions of the "action" for an identified Subject-Action-Object relation. AlchemyAPI normalizes actions by performing lemmatization, verb tense decoding, and negation detection.

How Do I Use It?

Verb action normalization is enabled automatically for all AlchemyAPI relations extraction API calls.

What Do Normalized Verb Actions Look Like?

The following is an example XML-formatted response (for "I should really consider buying an iPhone."):

<subject>
    <text>I</text>
</subject>
<action>
    <text>should really consider buying</text>
    <lemmatized>should really consider buy</lemmatized>
    <verb>
        <text>buy</text>
        <tense>future</tense>
    </verb>
</action>
<object>
    <text>an iPhone</text>
</object>

In the above example, lemmatization translates the verb "buying" into a base form of "buy". Additionally, AlchemyAPI identifies that this verb is being used in a future tense, as the relation discusses "considering buying" an iPhone.

NOTE: Verb action normalization is supported within all AlchemyAPI response formats, including XML, JSON, and RDF.

What about negation?

AlchemyAPI can detect negated actions, understanding the different between "buy" and "not buy". This is expressed in the following example (for "I should not consider buying an iPhone."):

<subject>
    <text>I</text>
</subject>
<action>
    <text>should not consider buying</text>
    <lemmatized>should not consider buy</lemmatized>
    <verb>
        <text>buy</text>
        <tense>future</tense>
        <negated>1</negated>
    </verb>
</action>
<object>
    <text>an iPhone</text>
</object>