My family research right now is based on digitalized 19th century vital records. These, of course, are scans of the original, not text. There is, though, a third party project where people have transcribed some of information, the form basically is a table where you can see that such and such was born/got married/died in year X, their father's (or on some rare occasions, mother's) name was such and the record number where further details can be found. 

So I copy-pasted information from parish of interest into spreadsheet, it has over 30 000 such records. Obviously I could just use this to look up information just on people I care about, however, to my dismay there are two large families with exact same surname, which so far don't appear to have any actual connection and things are further complicated by the fact that many people have same names. So I hatched this idea that since many of the records actually hold nice little piece of family tree (child-parent, husband-wife relationships, for females there could also be parent, husband, ex relationships indicated by surname change) these somehow could be stitched together by analyzing matches between people's names, relationships and lifetimes to give me probable family tree to verify, which would spare me from digging trough records on the other family, unless there is some well hidden relation. 

I am at loss how to do this. So far I've tried to convert data from spreadsheet to GEDCOM and make use of family tree software that finds duplicates, but it seems none of the applications have merging options that would be capable of doing such thing automatically, plus they look at exact data matches, not probable relationship based on years person was active and their relationships with other persons. I am wondering if anybody could suggest some new idea how to do this.