|
knoxcotn-digest Tuesday, April 24 2001 Volume 01 : Number 150
---------------------------------------------------------------------- Date: Tue, 24 Apr 2001 13:30:46 -0400 From: "Billie R. McNamara" <knox@tngenweb.org> Subject: [KnoxCoTN] Article about searching the Ellis Island records Forwarded with permission of Gary Mokotoff, the author of the article that appears below my editorial comments. This is the best real description of the Ellis Island site I've seen to-date. Gary's experiences pretty much mirror my own. Sir Spouse has a VERY unusual surname, and I couldn't even find his grandparents, whose entry record I have... The "similar" names list is also a little bizarre. I got maybe 5 on his surname. But, when I went through the soundexed film for just the reel his grandparents were on a few years ago, I found at leat 50 possibilities. A friend also suggested changing the beginning letter. A script "K" could've been misinterpreted as an "R"; a script "S" might've been read as an "L"; a funky "T" could've been seen as an "L"; etc. Sir Spouse's grandfather's given name was Istvan -- and it was clearly that on the microfilm I looked at -- but it's something like "Ishran" in the database. Not being able to get to the site has been frustrating; however, the biggest frustration for me has been that the system will occasionally let me in, but it will be "too busy" while in the middle of my search. Hopefully, they'll rectify that. I understand there've been some excellent suggestions from JewishGen and other sources on how to tweak the site. If I had a wish, it would be to be able to search for residence location. I'd love to know how many people from the tiny village from which my grandparents-in-law came migrated to the US during the period covered by the records.
>************************************************************ > Nu? What's New? > News About Jewish Genealogy > from Avotaynu > Vol. 2, No. 8 - April 22, 2001 > Gary Mokotoff, Editor >************************************************************ > >======================================== >Special Edition -- Ellis Island Database >======================================== > >Ellis Island Search Engine Has Great Shortcomings >================================================= >If your name has many spelling variants, some unknown to you, don't >expect to find your ancestors in the Ellis Island database. The >search engine used to retrieve records has great shortcomings. In >the last issue of "Nu? What's New?" I reported that there is no >soundex system. According to the Ellis Island Foundation, "a >computational linguistic search filtering system that is sensitive >to ethnic linguistic patterns" is an essential feature of the >system. Perhaps they should have stuck to plain old-fashioned >soundexing. > >With great dedication to my readership, I got up Sunday morning at >3 a.m. to access the database. Any other time of day on the East >Coast of the United States, attempting to reach the site generated >a system-busy notice, a problem not uncommon for popular >genealogical databases during their first few weeks of operation. > >I keyed in my family name, MOKOTOFF, knowing that there were >spelling variations of it used by my grandfather and his siblings >and parents. This produced three hits out of about 10 known records >in the database (Mokotoff is an uncommon name). The system then >allows you to select "Close Matches Only" or "Alternate Spellings >Only." (Only Exact Match is allowed in the initial search--a flaw >that should be corrected in some future enhancement). In both cases >there were no additional matches to the surname MOKOTOFF. > >I then tried the correct Polish spelling of my name, MOKOTOW, and >got only one hit, my grandaunt, Taube. The Closest Matches Only >option produced no additional hits. The Alternate Spellings Only >option also produced no hits. Still missing were the arrivals of my >grandfather, his brother Tuvia and his parents (frankly I have >never found the arrival of his parents--I had hoped the database >would find the record). Apparently MOKOTOFF is not considered a >variant of MOKOTOW. > >The next step was to try spelling variants. I deliberately >misspelled the name as MAKOTOW and up popped granduncle Tuvia, >spelled "Torvie Makotow." Testing the system's ability to catch >spelling variants, I clicked the Close Matches Only option and got >one hit: MAKOTOWITZ. I tried Alternate Spelling Only and got >nothing. Apparently MOKOTOW is not a variant of MAKOTOW in the >Ellis Island system. The most primitive rule of soundexing--drop >the vowels--is not implemented! > >It took me 15 years to find my maternal grandmother's Ellis Island >passenger arrival record because of the inadequacies of the Russell >soundex system--the standard soundex system of the U.S. government. >I knew her name was TARATOTSKY (Soundex code T633) but could not >find her in the microfilmed index to Ellis Island arrivals which is >organized in Russell soundex order. Only because I knew her year of >arrival and got lucky that she came to Ellis Island through the >port of Hamburg, Germany, was I able to find her in the Hamburg >Emigration lists as Frushe TARATOZKI (Soundex code T632). The >Russell soundex system does not equate the "ts" and "z" sounds--the >Daitch-Mokotoff soundex system, which I co-authored, does equate >the two. My genealogical research files were in my office, so, from >memory, I keyed in TARATOZKY and got no hits using searches Exact >Match, Close Match, or Alternate Spelling. I then recalled that >there was a list on my home computer of all persons named >Taratotsky (and variants) who came through Ellis Island. Locating >the file, I determined she arrived as TARATOZKI and immediately >found her in the Ellis Island database. Apparently their >"computational linguistic search filtering system that is sensitive >to ethnic linguistic patterns" does not equate names ending in -SKY >and -SKI. > >As noted above, there was only one case where I was able to >generate a Close Match. The name MAKOTOWITZ appeared as a Close >Match for MAKOTOW. The best guess is that Close Match means "Same >Initial Letters." > > >Misspellings in the Ellis Island Database >========================================= >To further confound the researcher, there is strong evidence that >there are many spelling errors in the database. My Mokotow family >came to Ellis Island from Warsaw, Poland, through German ports. All >records of their arrival show Warsaw with the German spelling: >WARSCHAU. The Ellis Island database shows that grandaunt Taube came >from WARSCHAN, and granduncle Tuvia came from WARSCHAW. On the >Taratotsky side of the family, Nahem Taratazky came from BRALYSTOK >(Bialystok), Leib Tartazky came from BIELISTOK, Chaim Tartazky came >from RIALYSTOK, Frusche Taratozki came from TALOWKI (Jalowka). > >There may be comparable errors in the names of the passengers. > > >The Importance of Soundexing >============================ >The importance of soundexing can be seen in my research into the >Ellis Island records for my maternal grandfather's surname: >TARATOTSKY. In my years of research, which demonstrated that all >persons with that surname from the region of Bialystok, Poland, are >related, I have found the following spelling variants in Ellis >Island records: Taratatski, Taratatzki, Taratatzky, Taratazki, >Taratazky, Taratoszky, Taratotzki, Taratozki, Taratzki, Taritazki, >Tarotazki, Tartacka, Tartacki, Tartacky, Tartatzki, Tartatzky, >Tartazki, Tartazky, Tartotsky. > >Using the Ellis Island database, it appears I would have to search >all 19 variants--plus some unknown number of variants due to errors >in data capture. Using the microfilm index that implements the >Russell soundex system, it is reduced to two searches. Soundex code >T632: Taratazki, Taratazky, Taratoszky, Taratozki, Taritazki, >Tarotazki, Tartacka, Tartacki, Tartacky, Tartazki, Tartazky. >Soundex code T633: Taratatski, Taratatzki, Taratatzky, Taratotzki, >Taratzki, Tartatzki, Tartatzky, Tartotsky. Using the Daitch- >Mokotoff Soundex System only one search would be required--all 19 >names have the same code: 393450. No wonder the Daitch-Mokotoff >system has been the standard for Jewish genealogical databases for >the past 15 years. For those that think it is a Jewish soundex >system or an Eastern European soundex system, I assure you the >Daitch-Mokotoff soundex system has applicability to any ethnic >group--but that is a subject for some future article in AVOTAYNU. >For those that are impatient, see my article "Soundexing Multi- >ethnic Computer Databases: A Proposed Solution". National >Genealogical Society Quarterly. Vol. 86, No. 2, June 1998. > > >Optional Searches >================= >In the Ellis Island system, once the initial search is performed, >there are a number of options solely of interest in those cases >where there are too many hits. You can qualify the search by >Gender, Year of Arrival (range of years), Ethnicity, Age of Arrival >(range of ages), Port of Departure (all ports for which there were >hits on the initial search), and Name of Ship (all ships for which >there were hits on the initial search). I searched for John Kennedy >and got 538 hits. I then qualified my search by Ethnicity and got >the following choices: American, British, Canadian, English, >French, Irish, Lithuanian, Newfoundlander, Russian, Scottish, >Unknown and White. Clearly Ethnicity refers to the Race column on >the ship's manifest. > > >Other Features >============== >Clicking the passenger's name provides greater detail--namely, >person's name, ethnicity, place of residence, date of arrival, age, >gender, marital status, ship of travel and port of departure. You >can then request the Ship Manifest, which lists all the passengers >on the page with the identified passenger. This list is browseable >providing the ability to go to the Previous Page and Next Page. >Finally, clicking the Ship link provides you with a picture of the >ship and its history. There is no evidence you can get a image of >the actual page from the ship's manifest. > > >Performance >=========== >The evidence is that once the usage of the system becomes more >normal, response time from the database will be excellent. At 3:00 >a.m., I got system-busy messages about 25% of the time. I had a >reason to jump back onto the Internet site at 6:00 a.m. to verify >some of the items discussed above, and I got very rapid responses >to my searches. By that time the Australians had gone to sleep and >the Americans had not yet awakened. > > >Cost >==== >The site is at http://www.ellisislandrecords.org. Use of the index >on the Internet is free. All items described above are displayed at >no charge. > > >Conclusion >========== >At least for Eastern European Jewish genealogy, with its multitude >of spelling variants, the index to the Ellis Island database is >wanting. With 19 spelling variants of Taratotsky for some 50 >passenger arrival records it is highly unlikely that any person >searching for that surname will find ancestors. > >My 35 years in the computer software business demonstrated to me >that if you get ever aspect of system design right but fail in one >and only one aspect, your end product is a low quality system. It >appears the Ellis Island database designers got every aspect right >but one: information retrieval. > > >************************************************************ >To be added or removed from this mailing list, go on >the Internet to http://www.incor.com/avotaynu.htm > >Back issues are available at http://www.avotaynu.com/nu.htm. > >To subscribe to AVOTAYNU, The International Review of Jewish >Genealogy, go to http://www.avotaynu.com/journal.htm. > >To order books from our catalog, go to >http://www.avotaynu.com/catalog.htm > >Copyright 2001, Avotaynu, Inc. All rights reserved > >Send requests to reproduce portions of "Nu? What's New?" to >info@avotaynu.com >************************************************************ ------------------------------ End of knoxcotn-digest V1 #150 ******************************
|