(Message inbox:86) Return-Path: <@hake.stanford.edu:gio@db.stanford.edu> Received: from Hake.Stanford.EDU by paris.ics.uci.edu id aa00466; 7 Jul 99 20:04 PDT Received: (from gio@localhost) by Hake.Stanford.EDU (8.8.8/8.8.8) id UAA16018; Wed, 7 Jul 1999 20:04:08 -0700 (PDT) Date: Wed, 7 Jul 99 20:04:07 PDT From: Gio Wiederhold To: Stephen Bay Cc: gio@db.stanford.edu Subject: Re: (DBWORLD) The UCI KDD Archive In-Reply-To: Your message of Thu, 1 Jul 1999 11:22:57 -0500 (CDT) Message-ID: Stephen The form is attached, it did not look like an on-line HTML/JAVA form, right? Feel free to FTP the files from www-db.stanford.edu/movies/* Gio >>>>>>>>>>>>>>>>>>> Guidelines for Documenting Data Sets: DATA SET INFORMATION The purpose of this page is to provide detailed information on a particular data set to enable other researchers to use the data for a variety of analysis tasks. For example, a data page might describe census data which could then be used for different analysis tasks such as classification or clustering. When filling out this form, simply place your answer after the point indicated by '>'. We will then process the form to ensure that all documentation files follow a common format. 1. Title of Data Set -- Indicate the central topic of the domain. > Movies 2. Data Type -- Indicate the type of data: multivariate, relational, time seris, sequential, images, spatial, text, time series, transactional, web data -- If the data is heterogenous, list all relevant types. Relational, with multi-valued fields and time-values. Suitable for objectification and inferencing novel and social relationships. A Remakes file is suitable for testing recursive processing. 3. Abstract A MAIN file listing over 10 000 films; with many older, odd, and cult films. Ancilliary files, useful for joining and inferencing on CASTS, ACTORS, PEOPLE, as directors and some producers etc, REMAKES, and some data on STUDIOS. Detailed descriptions of the fields and their formats is provided in a DOC file. The material includes some social information, as `lived-with' and `married to'. 4. Sources (a) Original owners of database Gio Wiederhold, Stanford University, 650 725-8363 www-db.stanford.edu/pub/movies/doc.html (b) Donor of database (name/snail address/phone/email/homepage) Gio Wiederhold, Stanford University, 650 725-8363 > www-db.stanford.edu/pub/movies/doc.html 5. Data Characteristics The original motivation was for database class exercises, to replace the boring `manager of the toy-department' queries. Note that the CASTS, refering MAIN and ACTORS is logically identical to the inventory file refering to suppliers and assemblies in the the standard bill-of-materials problems. Personal interests caused to be made complete for all Hitchcock movies and TV episodes. Related films by type and actor were added gradually. Subsequent research on temporal databases caused date fields (years only) to be added. It allows testing, say,if the dates-of-work of an ACTOR match the dates of the MAIN films that the CAST relation shiows. Object-oriented database features could be tested with fields having multiple and two-level values, as documented in DOC. The entries were gradually collected during course work starting about 1975 and are still being updated. Most of the entries were manual. The DOC file lists some of the reference works used. Corrections and additions continue to be appreciated. All Types: (a) Missing Values: -- Outside of key fields, missing values are common. Their encoding is described in DOC. Sometimes the data seems to be unavailable, sometimes it hasn't been entered. Some information, as `lived-with' is inherently incomplete. (b) Censored data: Minor actors are ignored. (c) Cost Information: (if applicable/available) -- none (d) Dependencies: yes, documented in detail in DOC. Every MAIN film must have a director in PEOPLE. About 50 pseudo director names ahve been listed in PEOPLE to allow interesting films to with (yet) unknown directors to be entered. Every CASTS entry must relate to a MAIN film entry. Every ACTOR should appear in some CASTS entry, but not vv. See DOC for more type information. Text: (a) language: Films are listed, if known, with their original language title. an Alt(T: ) field provides English translations, where known. (c) structured: the files are structured in multiple HTML tables, and can be directly displayed by capable browsers. The CASTS file brings many browsers to their knees. Image: some images of actors exist, but are not included because of copyright protection. 6. Other Relevant Information 7. Data Format The current files are in HTML, to allow easy parsing to other formats. An XML version is being considered. The approximate file sizes are DOC ....... 50K MAIN ...... 1 145K 11 400 entries PEOPLE .... 355K 3 290 entries CASTS ..... 4 340K 46 000 entries ACTORS .... 811K 6 800 entries REMAKES ... 135K 1 278 entries STUDIOS ... 26K 200 entries 8. Past Usage Class exercises, research test cases. Has included other universities, and some database companies. 9. Acknowledgements, Copyright Information, and Availability (a) copyright information Held by Gio Wiederhold, 1990, 1999 (b) usage restrictions No usage restrictions, other than for commercial resale. (c) citation requests Please acknowledge the source when used: Gio Wiederhold, Stanford University. (d) acknowledgements Many students, family, and friends who contributed entries. 10. References & Further Information (a) References on Movies are listed in DOC, but the objective is to have data that are easy to understand for a broad audience. (b) www-db.stanford.edu/pub/movies/doc.html (c) www-db.stanford.edu/pub/movies/doc.html ---------------------------------------------------------------------- /Gio Wiederhold/ http://www-db.stanford.edu/people/gio.html