I've added new functionality to the stylesheet that builds the serOgraphies.xml file. As part of the information it gathers on each character, it also pulls out the beginning of each speech (<said>), and tells you what the nearest <rs> to that speech is. In other words, it calculates which epithet is most closely associated with each speech. This involved writing a couple of useful functions (hcmc:getTextOffsetBetweenTags and hcmc:getNearestRsTag) which will also be useful with the Map of London.

It's currently set to a distance limit of 250 characters (meaning letters/digits/glyphs, not literary characters), so that if there is no <rs> within 250 characters of the <said>, no epithet is returned. That distance can be changed easily in the stylesheet.