PHOGLE is a database for photos, with a sophisticated system for tagging them, so that they can be found by search engines and browsed by categories.

PHOGLE currently contains photos by Erland Lewin, Mårten Stenius and Åsa Larsbo. We will be adding, rating and categorizing photos over time. Eventually we may open the system for other contributors.

All photographs in PHOGLE are copyrighted, and may not be used without permission.

We hope to establish a symbiotic relationship with search engines by providing well-tagged contents, and also with unobtrusive content-sensitive advertising programs like Google AdSense.

If you want to get in touch with us for any reason, send us an E-mail at webmaster@phogle.com.

Phogle licenses patented (U.S. Pat. No. 6,556,973) VoxiTM natural language technology from Voxi AB to make the semantic tagging system multilingual. This technology is in the process of being integrated into the site.

Phogle is an artistic expression, and as such subject to Swedish constitutional freedom of speech laws (Yttrandefrihetsgrundlagen and Tryckfrihetsförordningen). As such the site has a legally responsible publisher (Ansvarig Utgivare), who is this case is Erland Lewin.

Valid XHTML 1.0! Valid CSS!

June 17, 2005

Milestone 1 was achieved, with the basic functionality and a semantic browse function implemented, and working on Mozilla Firefox and Internet Explorer.

February 8, 2005

The phogle.com domain was registered, and the public interface became available.

January 30, 2005

Work on phogle's public interface began.

December 2003

The first work began on a personal database for my photos.

November 20, 2003

Mårten Stenius, Per Stenius and I bought a Minolta Dimage Scan Elite 5400 film scanner together, and the plans for digitizing my photo collecting could start being realized.

The goal of PHOGLE is twofold:

  • to enrich the world wide web by adding content in the form of photos with a rich set of multilingual, searchable keywords
  • to experiment with techniques for making web searching more successful.

Improving Search Technology

Pure text based searching has several problems PHOGLE would like to solve.

An E-type isn't an animal

If you have a picture with the keyword "Jaguar E-Type", and you search for "car", you won't get a hit. However, if you are looking for pictures of the animal jaguar, you would get a hit.

PHOGLE has a semantic database, with information like knowing that the E-type is a car, which will solve these problems by associating the keyword "car" with the photo. This information will also let you search specifically for the car model by using the site's browsing function.

But it is a voiture!

The most straight-forward issue in a multilingual world is to have keywords for a single 'tag' in many languages, so that a search for the french voiture or spanish carro and coche will also match the E-type.

...And is it rojo or roja?

However, multilinguality has a greater complexity which may not be evident to unilingual English speakers because English has very few conjugations of words.

In a language like Spanish, the adjective "red" can be written either rojo or roja depending on the gender of the word it is describing. So ideally you'd like to find the E-type regardless of whether you search for "rojo" or "roja", "car" or "cars".

When the technology is stable, we'd like to spin off the linguistic technology as an "Open Data" project to benefit all netizens.

PHOGLE is a hobby project, so the development efforts will come in uneven bursts.

Next Milestone (2): Multiuser

  • Let users log in
  • Create mechanism for uploading digital photos, with EXIF decoding
  • Let external users add photos to the database, and solve the legal issues associated with this.
  • Create a tagging tool
  • Let users create their own albums / slide shows with collections of photos

Milestone 3: Multilingual

  • Make the site generate language-specific versions of each web page.
  • Add a grammar module
  • Support one language other than english (probably Swedish)

Future work

This is a list of some ideas for the future, but it is in no way complete or in any specific order.

  • "a <profession>" when doing person to text with unknown name
  • Street address to text?
  • Bajada del nino info not shown properly for (see photo #122)
  • Rate all photos in the database
  • Tag all photos in the database
  • Scan all Erland's film negatives
  • Scan all Erland's slides
  • Add all Mårten's digital photos
  • Let users register
  • Let users rate photos
  • Let users comment on photos
  • Inheritable attributes (in semantics)
  • Include info from Wikipedia about the pictures
  • Show the names of deceased people
  • Semantics: Pinales can also be trees!
  • Investigate and improve cacheability of the images (can redirects be removed?)
  • Clean up typography
  • Create a favicon

9069 photos are in the database.

615 are fully tagged, and 6 are at least partially tagged.

Of the photographs rated by their photographers, this is the distribution of ratings:

  • Perfect: 15 (12 tagged)
  • Excellent: 7 (7 tagged)
  • Very good: 25 (25 tagged)
  • Good: 344 (294 tagged)
  • Average: 1464 (244 tagged)
  • Uninteresting: 1549 (21 tagged)
  • Snapshot: 1029 (5 tagged)
  • Poor: 261 (7 tagged)
  • What is it?: 27 (0 tagged)

All the photos on PHOGLE are copyrighted by their photographers and may not be used without permission.

We will be creating a price list for buying rights to use the photos, for now send an e-mail to webmaster@phogle.com if you want to use a photo.

Most photos are scanned at a fairly low resolution, but we can scan 35mm negatives and slides up to 5400 dpi on request.

Also note that most pictures have been scanned quickly in batch, and might benefit from being scanned with more care. Most images have not been color corrected etc.