Presentation is loading. Please wait.

Presentation is loading. Please wait.

Making Search Personal with NEPOMUK and Refinder

Similar presentations


Presentation on theme: "Making Search Personal with NEPOMUK and Refinder"— Presentation transcript:

1 Making Search Personal with NEPOMUK and Refinder
Dr Leo Sauermann, CEO getrefinder.com

2 Making Search Personal with NEPOMUK and Refinder
Hello, my Name is Can’t remember things Realizing the Semantic Desktop since 2002 Making Search Personal with NEPOMUK and Refinder

3 Brigitte Secretary of Prof. Andreas Dengel, DFKI KM Lab. Loves Skiing.

4 Making Search Personal with NEPOMUK and Refinder
personal search…. ok Making Search Personal with NEPOMUK and Refinder

5 Making Search Personal with NEPOMUK and Refinder
ok Brigitte Secretary of Prof. Andreas Dengel, DFKI KM Lab. Loves Skiing. Making Search Personal with NEPOMUK and Refinder

6 Secretary of Prof. Andreas Dengel, DFKI KM Lab.
Loves Skiing.

7

8 How are they related?

9

10

11 Alps ski Secretary of Prof. Andreas Dengel, DFKI KM Lab. Loves Skiing.

12 Alps ski

13 Alps ski

14 Leo Sauermann prototype: “Gnowsis”, 2003 http://leobard.net 28.11.2012
Making Search Personal with NEPOMUK and Refinder

15 Prof Stefan Decker Name: “Semantic Desktop” Idea: EU project
Name: “Semantic Desktop” Idea: EU project Making Search Personal with NEPOMUK and Refinder

16 Making Search Personal with NEPOMUK and Refinder
DFKI Ansgar Bernardi Andreas Dengel Team of artificial intelligence scientists Making Search Personal with NEPOMUK and Refinder

17 Making Search Personal with NEPOMUK and Refinder
Mandriva Linux Sebastian Trüg Stéphane Laurière KDE-Linux operating system Making Search Personal with NEPOMUK and Refinder

18 Making Search Personal with NEPOMUK and Refinder
Making Search Personal with NEPOMUK and Refinder

19 …to conduct research about the future of knowledge work, how to empower an individual knowledge worker to better exploit their personal information space and to maintain fruitful communication with others. Making Search Personal with NEPOMUK and Refinder

20 Making Search Personal with NEPOMUK and Refinder
Making Search Personal with NEPOMUK and Refinder

21 Desktop includes documents, images, music, videos, address book, personal calendar, todo lists, note-taking tools, s, … + future stuff people may need Making Search Personal with NEPOMUK and Refinder

22 Design Goals Make information reusable, Detached from the application
Link items, organize items Compare items, gather them in lists Use information in project/task context Automatic, based on artificial intelligence Cross-plattform on KDE-Linux, Windows, MacOsX establish open source semantic desktop standard Making Search Personal with NEPOMUK and Refinder

23 organize things using tags and stars
Making Search Personal with NEPOMUK and Refinder

24 Making Search Personal with NEPOMUK and Refinder
Example: Dolphin Assign tags, search for tags Making Search Personal with NEPOMUK and Refinder

25 Making Search Personal with NEPOMUK and Refinder
Example: Dolphin Assign tags, search for tags Making Search Personal with NEPOMUK and Refinder

26 Gwenview

27 plugin app file explorer browser extension api data base service data
Making Search Personal with NEPOMUK and Refinder

28 NEPOMUK Dazzler Blueprint
Making Search Personal with NEPOMUK and Refinder

29 Making Search Personal with NEPOMUK and Refinder
Making Search Personal with NEPOMUK and Refinder

30 plugin app file explorer browser extension api data base service data
Making Search Personal with NEPOMUK and Refinder

31 Works to integrate all data on the web.
Making Search Personal with NEPOMUK and Refinder

32 NIE Ontologies for rich data index
Create a rich index to build flexible applications on top Integrated & related “Contact” is used in s, im messages, author of documents, … Reusable Representation in RDF, XML, KEY/VALUE possible Standards-Based adaption of EXIF, vcard, vcal, MPEG 332 class types: contacts, calendaring, task management, downloads, file types, generic multimedia stuff, exif information for images, ID3 Tags for MP3s, messages and , file sharing, … personal information management: projects, tasks, organizations, places, …. 866 properties of these classes: structured data, a lot for calendar and multimedia formats. Making Search Personal with NEPOMUK and Refinder

33 RDF Database: Virtuoso
Stores all structured data and fulltext in a searchable index Fulltext Database Data Database RDF Graph Database Lorem ipsum… C S P O A file://x size 512 author nepomuk://hh-45 Lorem ipsum dolor… Making Search Personal with NEPOMUK and Refinder

34 Index the Data: crawlers
Initial indexing of the data into the store Strigi C++, Linux strigi.sourceforge.net Aperture Java aperture.sourceforge.net Making Search Personal with NEPOMUK and Refinder

35 Making Search Personal with NEPOMUK and Refinder
Making Search Personal with NEPOMUK and Refinder

36 Making Search Personal with NEPOMUK and Refinder
Improving the Indexer Initial indexing took 100% CPU time, little additional benefit over predecessor Alan Grimes, annoyed by 100% CPU time taken by indexer: ml Indexer takes less CPU and memory now. …What the fuck is this semantic bullshit? Computers most notoriously cannot think. They are purely storage and processing systems... Making Search Personal with NEPOMUK and Refinder

37 it’s the user experience, stupid
Multimedia and PIM use. More end-user applications. Making Search Personal with NEPOMUK and Refinder

38 Filter in the File Explorer
Multiple facets can be used Making Search Personal with NEPOMUK and Refinder

39 Semnotes http://smile.deri.ie/projects/semn/details 28.11.2012
Making Search Personal with NEPOMUK and Refinder

40 Semnotes Making Search Personal with NEPOMUK and Refinder

41 Making Search Personal with NEPOMUK and Refinder
Making Search Personal with NEPOMUK and Refinder

42 Making Search Personal with NEPOMUK and Refinder
Making Search Personal with NEPOMUK and Refinder

43 establish a semantic desktop standard community!
Making Search Personal with NEPOMUK and Refinder

44

45 Filter Big Bang Theory Episodes
Season 3, The Vengeance Formulation Making Search Personal with NEPOMUK and Refinder

46 Making Search Personal with NEPOMUK and Refinder
Filter Music ATB Albums, No Silence Making Search Personal with NEPOMUK and Refinder

47 Open files by tag Making Search Personal with NEPOMUK and Refinder

48 Save files with tag&description
Making Search Personal with NEPOMUK and Refinder

49 Making Search Personal with NEPOMUK and Refinder
Multimedia Use Cases Beyond TV Shows I'm using Nepomuk extensively everyday, …, to manage all my media, with the help of Nepoogle, Drop2TagIcon and Neposideckick, and results are amazing. Things like locate duplicate files or manage TVShows are only some samples of the capabilities of Nepomuk at desktop level. …I have 2133 video files … 35 are related to the actress 竹内結子 (Takeuchi Yūko), 11 files released in year 2012, and 3 have in common actor 阿部寛 (Abe Hiroshi). t=98468&sid=f90bf929b2f1faae2540e053 bab6f904&start=15#p221521 Synonyms, multi-linguality Relations between files and actors Actors and their relations are copied from web-databases and stored on semantic desktop to enhance local queries Making Search Personal with NEPOMUK and Refinder

50 Making Search Personal with NEPOMUK and Refinder
Zeitgeist Plans to integrate Zeitgeist and NEPOMUK for KDE Making Search Personal with NEPOMUK and Refinder

51 Finding duplicate files
select distinct ?u1 ?u2 where { ?f1 a nexif:Photo . ?f2 a nexif:Photo . ?f1 nfo:hasHash ?h . ?f2 nfo:hasHash ?h . ?f1 nie:url ?u1 . ?f2 nie:url ?u2 . filter(?f1!=?f2) . } Making Search Personal with NEPOMUK and Refinder

52 Filter & Drill down on data
Virtuoso RDF Store SELECT * WHERE { ?s nco:fullname ?name . ?name bif:contains "'brigitte*'". } Brigitte Structured queries Fulltext queries Inference Engine: rules, ontologies, generates new facts Making Search Personal with NEPOMUK and Refinder

53 Answer complex questions
Through natural language processing, full-text analysis and the index of relations “Whats the phone number of the KM Group Secretary?”  Not possible to answer using fulltext search. Only SIRI or Watson have the superpowers to… Making Search Personal with NEPOMUK and Refinder

54 Map the query to the graph
The answer is found by spreading activation and partial queries Property <phone_number> Class <secretary> Instance <KM-Group> (no result) <KM-Group><unit><SecretaryAD> <SecretaryAD><phone_number> „ “ [1] [2] [3] [4] instances: <SecretaryAD>, … “phone number of the KM-Group secretary” [5] Combining Metadata and Document Search with Spreading Activation for Semantic Desktop Search, Schuhmacher, K.; Sintek, M. & Sauermann, L. ESWC 2008 Making Search Personal with NEPOMUK and Refinder

55 PIMO to build a personal network
Project Meeting Location Concetps & Associations to organize Connected with documents Inspired by basic human concepts Kaiserslautern Andreas Kickoff nepomuk Brigitte Documents Papers Projects nepomuk Research Files s Inbox Todo SAP nepo Karlsruhe Contacts Andreas Dengel Brigitte Selzer Sven Schwarz Associations are like a human person thinks Semantics give meaning to the elements in a computer Lets focus on a detail here. Relations with documents Individual s can be categorized to a project PIMO - a Framework for Representing Personal Information Models

56 Supports Orienteering Behavior
“users go somewhere, look around, know where they are, go on” Filtering keeping orientation Long-term users stopped using web-search interface spatial Teevan, J.; Alvarado, C.; Ackerman, M.S. & Karger, D.R. The perfect search engine is not enough: a study of orienteering behavior in directed search CHI 2004, pp Semantic Desktop for PIM

57 Making Search Personal with NEPOMUK and Refinder
NOKIA Search, on the phone „Brings the power of computers to mobile devices“ N900 and N810 are Maemo Linux-powered Nokia M9 Phone, millions shipped Tracker indexing framework maemo.nokia.com Making Search Personal with NEPOMUK and Refinder

58 Making Search Personal with NEPOMUK and Refinder
Philip Van Hoof, Last week inter-app information sharing on a high level using a inter-app defined and shared ontology with a query language suitable for the purpose, SPARQL, combined with a efficient file system meta data indexer. We delivered that and proof that its high quality software exists in the fact that a few million N9 users, yes a few million, are using it right now and yet there are no massive protests on forums about corrupted data: Tracker's SPARQL endpoint is today tested and robust high quality software. I salute Jürg [Billeter] for that. mail.gnome.org/archives/tracker-list/2012-November/msg00009.html Making Search Personal with NEPOMUK and Refinder

59 Making Search Personal with NEPOMUK and Refinder
Nepomuk, in the cloud Making Search Personal with NEPOMUK and Refinder

60 Connect and Crawl + Connectors: Scanning of any cloud app.
Dropbox Evernote Contacts RSS Twitter Docs Basecamp Box.net SkyDrive Campfire MindMeister Calendar + November 2012

61 Refinder Cloud Search Scalable, Full-text, Sorting, Faceting, Filtering, Highlighting, customizable Ranking. Scalable full-text and phrase search and/or term conjunction Sorting and faceting on any item attribute Customizable relevancy ranking Results including item attributes, result highlighting, previews, and filter options November 2012

62 Refinder Cloud Search More cloud apps, growing data, small screenspace Desktop Data  Personal Data Desktop Apps  Cloud Apps November 2012

63 Making Search Personal with NEPOMUK and Refinder
Timeline of NEPOMUK and adoption rates developers users 4 guys 3k Refinder ~100s 0.6 mio Nokia maemo & MeeGo ~10 25k Aperture ~10 million NEPOMUK & Gnome ~10 millions NEPOMUK-KDE 4.0 ~60 100 EU-IST Research Making Search Personal with NEPOMUK and Refinder

64 Making Search Personal with NEPOMUK and Refinder
Join! nepomuk-kde mailinglist Making Search Personal with NEPOMUK and Refinder

65 Making Search Personal with NEPOMUK and Refinder
I am looking for a partner to scale Refinder up nepomuk-kde mailinglist Making Search Personal with NEPOMUK and Refinder


Download ppt "Making Search Personal with NEPOMUK and Refinder"

Similar presentations


Ads by Google