In https://en.wikipedia.org/api/rest_v1/page/summary/Frederick_Chilton the returned extracts are:
"extract":"Dr.","extract_html":"<p><b>Dr."
In https://en.wikipedia.org/api/rest_v1/page/summary/Frederick_Chilton the returned extracts are:
"extract":"Dr.","extract_html":"<p><b>Dr."
Thanks for the report. The issue is with TextExtracts Reading infrastructure are currently in the process of deploying a rewrite of the summary logic.
Using that I see this is fixed with the following extract:
"extract": "Dr. Frederick Chilton is a fictional character appearing in Thomas Harris' novels Red Dragon and The Silence of the Lambs.", "extract_html": "<p><b>Dr. Frederick Chilton</b> is a fictional character appearing in Thomas Harris' novels <i>Red Dragon</i> and <i>The Silence of the Lambs</i>.</p>" }
I've merged this into the epic task which mentions this problem.
The fundamental difference that solves this problem is that we now use the first paragraph (HTML p tag) (the intro) rather than the TextExtracts approach of trying to count and pick out sentences in the preview by regex.