Sometimes the "summary" property of articles is just garbage. After a more precise fulltext is achieved through #10 , use algorithms like node-nlpsum to generate a good snippet consisting of a few important sentences.
Whenever a good enough fulltext-length is given, this way should be used instead of relying on meta-properties like open graph stuff.
Sometimes the "summary" property of articles is just garbage. After a more precise fulltext is achieved through #10 , use algorithms like node-nlpsum to generate a good snippet consisting of a few important sentences.
Whenever a good enough fulltext-length is given, this way should be used instead of relying on meta-properties like open graph stuff.