How to get from there to here, in my search engine? Well, once again, the simplicity of my site is the overiding consideration. There are two things we (not the royal ‘we’, I’m just assuming I’ll have some help) are archiving are oral histories and church data. Of course, if there was enough scholarship we listed, that would be searchable. At first, I imagine there will be a limited number of books (or titles, if permission not given to make the whole work available) and these will be listed and not really be ‘searched’ that often. However, it is conceivable that there will be enough works on there to make a quick scan a pain and people might want to only search those works that reference a particular topic, so forget what I just said, there are three things that are searchable. But they are very simply searchable. You would be searching the oral histories, I suppose, for categories like location, or topics like church or employment. The church records are already going to be organized in categories (denomination, attendance, etc.) that make searching mostly superflous. Thus, I am not sure that much more than the simple old-school Alta Vista method (I just love old-school, I can’t help it; bring back 45s, the board game, the test pattern, uh, pardon me) is all I really need for a reasonably decent search. I’m not sure how I would do a page rank type of thing, if any one has any ideas let me know. I think locale, like Michigan or West Virginia, would be of importance to people searching, as would oral histories dealing with type of employment like farming or mining, as might more specific areas (for our customers) such as social and church life. Perhaps family ties, marriage, etc. would be a very searchable commodity. How to weight these to appear when people put in those search times other than putting those histories that mention them most first I do not know. Now that I have some inkling of how to do an interior search engine I think it will be easy for this site. The data on our site will be categorized so that visitors can go right to the area they wish to peruse. This does not mean they will not search, however, and a search engine will still be useful. A visitor might be looking for something very specific, perhaps preachers, or a term too specific for our categories. It should be easy enought to accomodate those situations. It makes sense to utilize location as an important factor. As a full text may mention preacher or farming or any such term many times, it makes sense that a better match is in the metadata The church records will be in tables that the metada will list by denomination and type so if a person is just searching for Baptist, the category on the home page will take care of that. It is in the searches for a particular church or particular are of the country, that the search engine will be most useful. Thus, it seems logical that we give five points for a metadata hit, and one point for a full text hit. The oral histories have no titles, save a person’s name, but the first sentence of a paragraph should carry greater weight than when mentioned anywhere in the text. We will give three points for any hit in the first sentence of a paragraph. Oral histories are one of those rare occasions where randomness may be appropriate. There are so many histories, and visitors may benefit from having a few random to see what they are like.
You can’t get there from here
Advertisement
