Topic Maps – semantics for humans? WNRI Seminars on Semantic Technologies, 2010-12-15 Lars Marius Garshol <email@example.com> http://twitter.com/larsga
Agenda What are semantic technologies? Introduction to Topic Maps Topic Maps and classification A short history of Topic Maps The standards Topic Maps and RDF Example applications Software Learn more
Semantics? Semantics the study of meaning (orig. the meaning of words) Semantic technologies describe not just data, but also the meaning of data in traditional technology meaning is only in code and human interpretation John Searle, "The Chinese Room"
Non-semantic data What is this? How many entities are represented here? What is entities and what is properties?
The schema People often say that the schema defines the semantics But it's not really very semantic, is it? XML is not a semantic technology
Topic Maps example Separates entities from properties Relations are clearly visible We know the names of all entities Can query for all 人 and get all instances The full meaning remains obscure Types 人 subtype subtype 男性 女性 出来事 type-instance type-instance 源氏 夕顔 参加 参加 夕顔との出会い
Semantic technology Far richer description of concepts arbitrarily complex description of classes and properties Vocabularies can be reused across applications Data can be automatically merged Some of the meaning in the data can be modelled
Semantic technologies Topic Maps ISO standard, much used in portals emphasis on "human" semantics RDF W3C standard, foundation of the "semantic web" heavy use of logic in the stack of standards Other alternatives many other technologies want to be seen as semantic; how many of them are is disputable only widely-accepted standards really matter
What are Topic Maps? Uses for Topic Maps Introduction
What is Topic Maps? A technology for knowledge integration describes concepts and their relations allows documents to be attached to the concepts concepts can be matched across different topic maps matching allows topic maps to be merged seamlessly
What can Topic Maps be used for? Primary usage organizing information so you can find what you are looking for common example: portal or intranet less common: online publishing However, Topic Maps is really just a way to organize information can therefore be used for nearly anything Other uses e-learning real knowledge management decision support systems ...
From documents to topics The TAO of Topic Maps How to make a topic map
The Topic Maps approach (index) (content) topic map documents Create a conceptual map of the information being organized concepts and relations connections to documents (landscape) Like a book with an index or landscape and a map
Creating a topic map Analyze the documents Select the key concepts (topics) Analyze the key concepts (topic types) Identify their relationships (associations) For each topic, connect relevant documents (occurrences) Voila!
1. Document analysis Key concepts What is it? Evaluation report from the MODE project MODE, (Evaluation) CV of Jane Doe Jane Doe, (CV) Budget for IT group IT group, (Budget)
2.-3. Topics, with types Person Department Project Jane Doe IT group MODE
4. Adding associations employed in worked on part of worked on Consumer products part of employed in Documentation Roger Roe Jane Doe IT group MODE
5. Adding occurrences Jane Doe CV budget evaluation worked on employed in IT group MODE
6. The TAO of Topic Maps worked on MODE Jane Doe Topics represent things of interest Associations represent relations between topics Occurrences connect topics to information resources with relevant information
How to find information? Metadata as solution Metadata as problem Metadata
Metadata The obvious solution to the problem is to describe the documents that is, to attach metadata to the documents metadata in this context is “information about a document” So how does this help? it’s useful for managing the content it provides a better starting point for search it means better search results can be displayed it helps the user determine whether or not a search hit is interesting But is it what the user is looking for? the user starts out wanting to know more about a subject traditional metadata, however, focuses on the document if aboutness is provided at all, it gets squeezed into a single field Title: Recurrent Herpes Simplex Sciatica and its Treatment with Amantadine Hydro... Author: D.A. Fisher Date: 1982-05 Format: text/html Keywords: sciatic neuralgia, aman...
What’s wrong with keywords? The main problem is that their use is uncontrolled This leads to problems like authors misspelling keywords, authors using different keywords for the same thing, and authors using keywords that make no sense A secondary problem is that short of guessing, there is no way for the user to find out what keywords have been used The main benefit is that it’s cheap and simple
Taking control over the vocabulary The obvious solution is to create a list of legal keywords this is what’s known as a controlled vocabulary in a controlled vocabulary keywords are called terms this requires somewhere to keep the list, and a process for adding new terms Benefits gets rid of the misspelling problem gets rid of the problem with authors using different terms for the same thing Disadvantages introduces some overhead a flat list is difficult to manage users can still search using the wrong terms users will still have difficulty finding terms if the list is long authors will have the same problem
Organizing the terms The solution is clearly to organize the terms somehow In one sense we’re now back to the problem we had originally with documents the solution is also the same: we need to describe the terms somehow the difficulty is: what can you say about terms? The good news is that there are many traditional and well-known ways to approach this
Two worlds amantadine hydrochloride sciatic neuralgia Title: Recurrent Herpes Simplex Sciatica and its Treatment with Amantadine Hydro... Author: D.A. Fisher Date: 1982-05 Format: text/html Keywords: sciatic neuralgia, amantadine hydrochloride ? ? ? Metadata Subject-based classification
Describing the terms Tags Taxonomies Thesauri Classification approaches
Subject-based classification There are many possible organizing principles for documents By author time period genre etc Subject-based classification classifies documents by their subject the subject is what the document is about that is, the subject matter of the document Subject-based classification does not have any particular structure it's just an approach, and there are many different ways to do it
Folksonomies and tags Tags have recently become popular on the web used by web 2.0 sites like Flickr, Technorati, del.icio.us, ... also much used in blogs to categorize the posts Tags are effectively a controlled vocabulary of keywords except the control is often extremely lax The same benefits and problems del.icio.us for example has tags like xtm, topic_maps, topicmaps, topic_map, and topicmap
Taxonomies BT Organizes the keywords into a tree the most general at the top, more specific as you go down common structure used by Yahoo!, LOS, Dewey classification... Requires relationships between terms the relationships state that one term is more specific than another http://www.dmoz.org
A taxonomy example Nervous system disease Autonomous nervous system disease Peripheral nervous system disease Cauda equina syndrome Diabethic neuropathy Sciatic neuralgia
Thesauri USE BT BT RT SN Thesaurus Taxonomy Folksonomy An extension of taxonomies come from the library world; much used in publishing the main extension is that thesauri add more relationships What thesauri contain: BT the same relationship as in taxonomies RT related term, which goes across the hierarchy USE refers to a term that should be used instead of the current one SN scope note, a definition of the term
A thesaurus example Nervous system disease Autonomous nervous system disease USE Peripheral neuropathy Peripheral nervous system disease Cauda equina syndrome Diabetic complications Diabetic neuropathy Sciatic neuralgia RT
Faceted classification The term “faceted classification” has been used to mean many different things originally invented by S. R. Ranganathan in the 1930s Faceted classification defines a number of facets or dimensions defines a set of terms within each facet sometimes these terms are arranged in a taxonomy documents are classified against each facet separately
Colon Classification Ranganathan's original faceted classification system Consisted of five facets: Personality The main subject of the document Matter The material or substance the document deals with Energy The processes or activities described Space The location described Time The time period described This has sometimes been referred to as “PMEST”
An example of use The Norwegian wine monopoly describes its products using these facets: type: red wine, white wine, beer, ... country of origin: France, Norway, ... price matches food: pasta, cheese, fish, beef, ... bottle size
Ontology in Topic Maps A Topic Maps model of some specific aspect of the world Worked on MODE Project Person CV Jane Doe ontology instances worked on CV
Taxonomies and thesauri revisited From the Topic Maps perspective taxonomies are an ontology terms become topics (of type “term” or “concept”) relations become associations (of various types) scope notes become occurrences However, in Topic Maps it’s possible to be more precise Nervous system disease Autonomous nervous system disease USE Peripheral nervous system disease Peripheral neuropathy Body part Cauda equina syndrome Disease Diabetic neuropathy Drug Amantadine hydrochloride Sciatic neuralgia Peripheral neuropathy Part of Attacks Treats
Expressivity progression Topic Maps Taxonomies, thesauri Flat list, tags Expressivity No model Closed model Open model
Metadata revisited Metadata can also be represented in Topic Maps create topics for the documents map fields to names, occurrences, or associations Big pharma Amantadine hydrochlorine Sciatic neuralgia attacks about author of Peripheral nervous system treated by D.A. Fisher This part is untrue! produced by works for Recurrent Herpes Simplex and its... Date: 1982-05 Format: text/html
Benefits of Topic Maps Richer, more expressive model multiple paths to the information you seek typed associations provide “signposts” along the path Improved support for search search for concepts, rather than just documents associations can be used for filtering Merges classification and metadata into a single model greater expressivity (again) simpler architecture: just one system to relate to Maps directly to web portals easy to build and maintain web portal based on the topic map
Conclusion Traditional findability solutions metadata: describes documents classifications: gather and loosely organize keywords/terms Traditional solutions focus on documents Users focus on subjects Topic Maps open model for describing anything focus on subjects easily supports both metadata and existing classifications
What it actually looks like Deeper into Topic Maps
Advanced concepts Association roles Reification Scope Identity
Associations have no direction Puccini Angeloni pupil of
Instead associations have roles Puccini Angeloni pupil of pupil teacher
Richer relationships father child Lars Marius Bjørg Knut parenthood mother
Roles, role players and role types person pupil teacher teacher-of person topic type role type role type association type topic type association role player role player role role puccini angeloni N.B. role == association role and role type == association role type
Symmetric relationships country neighbor neighbor borders-with country topic type role type role type association type topic type association role player role player role role norway sweden
Reification From latin “re” = “thing” i.e. “thingification” In Topic Maps, for “thing” read “topic” So reification is about turning something into a topic Specifically it is about turning topic map constructs that are not already topics (i.e., names, occurrences, associations, association roles, and topic maps) into topic Useful for annotation of Topic Maps constructs
Reification example Ontopia start date 2000 2007 LMG's employment end date employed by Lars Marius Garshol Obviously, this is no longer the case. But how can we express that?
The semantics of reification Many possible interpretations of what the reifying topic represents: the same thing as the association the association as Topic Maps construct the assertion of this particular association Topic Maps reification is case (a) RDF reification is not formally defined, but is case (c)
Scope Every statement in a topic map has a scope that is, a set of topics representing the context in which the statement is valid the empty set is known as "the unconstrained scope" Abugida Alphasyllabary Bright Daniels Tibetan script
Applications of scope Multilinguality scope names and occurrences with language topics Authority scope statements with the authority that supports them Provenance scope statements with their source Time scope statements with the era in which they were true
Multiple topics in scope The context is the intersection of the topics a statement scoped with "Thursdays" and "LMG" is true on Thursdays according to LMG Implication: adding topics to the scope narrows the context of validity given a statement s in scope a, and s' in scope a, b we can see that s' is actually redundant
Topics and subjects A topic is a representation of a subject topic: Topic Maps construct representing subject subject: real-world thing subject topic Patrick Durusau "subject: anything whatsoever, regardless of whether it exists or has any other specific characteristics, about which anything whatsoever may be asserted by any means whatsoever" --ISO/IEC 13250-2:2006
Subject identification Topics can have globally unique identifiers attached to them these identifiers really identify the subject of the topic, and not the topic itself the identifiers are URIs However, these are of two different kinds...
Subject locators A subject locator is a URI that points to the information resource which is the subject Patrick Durusau depicted-in Photo of Patrick taken-at Leipzig http://larsga.geirove.org/photoserv.fcgi?t121182 subject locator http://larsga.geirove.org/photoserv.fcgi?t121182 same as URI of photo
Subject identifiers A subject identifier is a URI which refers to an information resource describing the subject Patrick Durusau depicted-in http://psi.ontopedia.net/Patrick_Durusau Photo of Patrick
Merging In Topic Maps, two topics must be merged if they have the same subject identifier, subject locator, or reified construct The rationale is that if this is the case they must represent the same subject
Example Patrick Durusau depicted-in Photo of Patrick http://psi.ontopedia.net/Patrick_Durusau taken-at Leipzig editor-of ISO/IEC 13250-5 Patrick Durusau editor-of http://psi.ontopedia.net/Patrick_Durusau ODF
Example editor-of ISO/IEC 13250-5 Patrick Durusau depicted-in editor-of Photo of Patrick http://psi.ontopedia.net/Patrick_Durusau taken-at ODF Leipzig
On merging Merging is not a special operation happens every time Topic Maps data is loaded Allows exchange of fragments identifiers ensure that fragments are reassembled simply by being loaded Allows reuse of data define identifiers for vocabulary (pieces of ontology) or for individual entities
Examples of use Subclassing SIs for this are defined in the standard can be interchanged between tools Hierarchy definition SIs for this were defined years ago; widely used today Schema language SIs defined in TMCL (about which more later) Countries and languages SIs defined by OASIS ...
LOS A common classification for public information in Norway published by Norge.no (Norway.no) http://norge.no/los/ Consists of a taxonomy of subjects, a taxonomy of geographic locations, and a set of classified resources Defines PSIs for the subjects and locations Used by Bergen Kommune
Grep The Norwegian National Curriculum basically the official definition of what children should learn in school published as a topic map by the Ministry of Education uses PSIs for all elements Currently starting to be used NRK project used it others are connecting to it, too an aggregator service is being built
Linked Open Data? This is linked open data using URIs to automatically connect statements from disparate sources Represented in different ways some use RDF some use Topic Maps and some, probably, use other things Called "Global Knowledge Federation" in the TM community the concept remains the same interchange across technologies is possible
HyTime The Davenport project ISO A bit of history
HyTime An ISO standard for hypertext first published as ISO/IEC 10744:1992 very ambitious and complex based on SGML (precursor of XML) many kinds of hyperlinks including links with any number of anchors, where each anchor is associated with a role type specifying its meaning... contains a metamodel for representing content to allow detailed addressing into any form of resource ...
Small beginnings 1991 The Davenport Group: project to merge back-of-book indexes to UNIX documentation from different publishers First attempt known as SOFABED (failed) 1993 CApH was set up, to use HyTime to solve the problem turned SOFABED into Topic Navigation Maps 1996 picked up by ISO committee responsible for SGML
ISO and TopicMaps.Org 1998 Topic Maps standard submitted for final ballot an SGML architectural form based on HyTime SGML syntax today known as HyTM W3C publishes XML 2000 ISO publishes ISO/IEC 13250:2000 (still in SGML) TopicMaps.Org created to produce an XML version of Topic Maps 2001 XTM 1.0 published by TopicMaps.Org in March
ISO 2001 work begins on data models for Topic Maps an infoset-based model, close to XTM 1.0 a graph-based model, far more abstract lots of politics, holding up all other work first commercial engine released (Ontopia) 2002 ISO publishes ISO/IEC 13250:2002 (with XTM 1.0) the first Norwegian portals start appearing 2006 ISO publishes ISO/IEC 13250-2:2006 – Topic Maps – Data Model
A little ISO history Topic Maps Data Model The Topic Maps Standards
The new ISO 13250 A multi-part standard, consisting of Part 1: Overview of Basic Concepts Part 2: Data Model Part 3: XTM syntax Part 4: Canonical XTM Part 5: Reference Model Part 6: Compact Syntax Part 7: Graphical Notation
Roadmap to the TM standards ISO 18048 QUERY LANGUAGETMQL ISO 13250 XTM SYNTAX CXTM SPEC CTM SYNTAX GTM NOTATION ISO 19756 CONSTRAINT LANGUAGETMCL DATA MODELTMDM REFERENCE MODELTMRM
The Topic Maps Data Model (TMDM) Created to define meaning and structure of topic maps Syntaxes map to this structure, as do TMQL and TMCL Defines the meaning of topic map concepts using prose Defines “subject”, “topic”, “scope”, “association”, ... Defines their structure using the information set model Just like XML Infoset Describes the kinds of things that exist in topic maps, and their properties Adds constraints on the model Rules for allowed values Also defines when merging happens, and how
How TMDM works One information item type defined for each topic map construct Complete list shown below One set of properties defined for each construct Example below: all topic map objects have item identifiers
Association Associations have the following properties: [type]: topic defining the association type [scope]: set of topics making up the scope of the association [roles]: set of association role items [reifier]: topic reifying the association [source locators]: URIs pointing back to element(s) the association came from [parent]: the topic map Merge if equal values for [type], [scope] & [roles]
Merging rules in TMDM One merging rule defined for each information type Equality rule says which properties to compare (as for association) Merging rule says how to merge two equal information items For topics, the equality rule is that two topics are equal if same value in [subject identifiers] property of both, or same value in [subject locators] property of both, or same value in [source locators] property of both, or some extra conditions Merging topics is done by creating a new topic item, whose properties contain the union of the old values, then replacing all occurrences of the old items throughout the model with the new one
Things A thing in the real world S A symbol in the computer domain The heart of RDF and Topic Maps is the same: symbols representing real-world things Both RDF and Topic Maps consist of statements about these things
Technical comparison Topic Maps and RDF are graph-based data models, have well-defined identity tests and merging operators, have XML-based interchange syntaxes (as well as human-friendly ones), are standards, and have standardized schema and query languages Differences RDF is lower-level than Topic Maps, Topic Maps support reification, complex context, and n-ary relationships, and Topic Maps distinguish different kinds of URI references
Timeline MCF-XML RDF Schema PICS-NG MCF RDF WD OWL RDF Rec '91 '92 '93 '94 '95 '96 '97 '98 '99 '00 '01 '02 '03 '04 ISO work starts XTM to ISO Standard finished ISO 13250:2003 SOFABED model ISO 13250:2000 XTM 1.0 Davenport Group TopicMaps.Org Topic navigation maps
Assertions RDF has one kind of assertion: the statement subject, predicate, object Topic maps have three kinds (1) Names (2) Occurrences (3) Associations “...” “...” http://www...
Handling of identity Topic Maps subject locator subject identifier item identifier RDF uri blank node The distinction between a URI referring to a description of the subject, and a URI referring to the subject cannot be expressed in RDF.
TMCL vs RDFS/OWL TMCL schema language validation semantics only very little reasoning or logic designed to support validation and introspection RDFS/OWL ontology description languages reasoning semantics only strong basis in logic OWL is essentially Description Logic
Semantic Portals eLearning Business Process Modelling Product Configuration Information Integration Metadata Management Business Rules Management IT Asset Management Asset Management (Manufacturing) ... Applications of Topic Maps
forskning.no Norwegian government portal to popular science and research information basically an online popular science journal owned by the Norwegian Research Council Purpose: To present science and research information to young adults Intended to raise interest and recruitment
Content of forskning.no The main content is articles about science and research subjects There is also a classification system used as a navigational structure The site is entirely topic map-driven Navigation structure is a topic map Articles are represented as topics Even images are topics...
Medicine Science Odontology Human body Volcanoes Clinical Med. Hormones The Brain Neurology Oncology The Dual Classification
The subject Subjects Fields People Articles A Subject
Article Subjects Fields Next article People An Article
Person Title Home page Mentioned in Employer A Person
The Project Wide ontology; research covers everything Ontology was created by reusing an existing thesaurus, automatically converted A series of 4-5 workshops established the basic principles Finally, the publishing application was built by Bouvet software used is ZTM (Python-based, open source)
Maintenance Maintained by central editorial staff in Oslo Articles written by distributed network of authors Authors write and submit articles online Articles enter workflow and are added by editors Editors also add connections to topic map
City of Bergen Second biggest city in Norway 250,000 inhabitants and 20,000 employees spends roughly 3 million USD annually on the portal project goal: to make all city services available through the portal Strong technology platform Oracle Portal + Oracle RDBMS Escenic as CMS Ontopia as Topic Maps engine DB2TM for data integration
Bergen: who does what? Most of the site is produced by Ontopia Some parts by Escenic Some are independent And some are service-specific portlets Static Escenic
Bergen architecture Service Catalog Oracle Portal Fellesdata Ontopia Dexter DB2TM TMSync Agresso Escenic Ontopoly LOS Editors
NRK/Skole Norwegian National Broadcasting (NRK) media resources from the archives published for use in schools integrated with the National Curriculum In production opened late 2008 Technologies Ontopia DB2TM conversion MySQL database Tomcat application server
Curriculum-based browsing (1) Curriculum Social studies High school
One video (prime minister’s husband) Metadata Subject Person Related clips Description
GREP Norwegian national curriculum published as a topic map has global IDs on all topics NRK/Skole clips attached to knowledge goals global IDs are in the topic map Therefore... Grade Subject Section Goal GREP Clip NRK/Skole
ndla.no Portal organizing learning resources into the curriculum to be integrated with NRK/Skole
Using Ontopia DB2TM converts to Topic Maps a simple XML mapping file this is enough to provide full sync Generic SDshare implementation listens for change events produces corresponding feeds ERP DB2TM Ontopia SDshare Server
Hafslund – points to note Extremely loose coupling ontology can be freely changed Very simple integration in many cases just an XML configuration file Very flexible architecture adding new sources is trivial Has more uses than just archiving once the data is collected...
E-learning Topic maps are associative knowledge structures They reflect how people acquire and retain knowledge BrainBank is used by students to describe what they have learned Initial users are 11-13 year olds who haveno idea what a topic map is… They capture the key concepts, name them, describe them, and associate them with others This helps them Capture the essence, Describe what they have learned, Keep track of their knowledge, and Lets the teacher help them BrainBank was built using Ontopia An application of the Web Editor Framework Demonstrates user-friendliness of TM editing
Business process modelling A multinational petrochemical company uses Ontopia for managing business process models The flexibility of the Topic Maps model allows arbitrary relationships to be captured easily Processes are modelled in terms of The steps involved, their preconditions, their successors, etc Processes can be related through Composition (one process is part of another), Sequencing (one process is followed by another), Specialization (one process is a special caseof a more general process)
Product configuration A Scandinavian telecom company uses Ontopia to manage product configuration Products belong to families Features belong to either products or product families Features are grouped in feature sets There are dependencies between features etc. The system models dependencies in a topic map Product configuration engineers use this to configure products using a user-friendly interface After each change, interface gives feedback on whether selection was valid Features Product families Versioning System data Products
Product configuration (2) Feature 1 The features are arranged in a tree trees vary in size (700-2500 features) two kinds of parent-child relationships (mandatory or optional) Configuration rules run across three different kinds of rules expressed as associations In addition: variables these have different values for different products Feature 2 conflicts-with requires Feature 3 Feature 4 Feature 5
Product configuration (3) The network of dependencies is already quite complex Now throw versioning into the mix! Managing all this data is not easy The system is driven by inference rules These work on the topic map Easily capture complex logic Also integrates with product documentation Very complex topic map at the last count ~20,000 topics and ~1,000,000 associations running complex queries on this really exercises the query engine
Business rules management (1) The US Department of Energy has used Ontopia to manage guidance rules for security classification Information about the production of nuclear weapons is subject to thousands of rules Rules are published in 100s of documents Most documents are derived from more general documents
Business rules management (2) Guidance topics form a complex web of relationships that is captured in a topic map Concepts are connected to if-then-else rules This constitutes a knowledge base (KB) KB used with an inference engine to automatically classify information (documents, emails, ...), and redact information (PDF, email, ...) Benefits: Model expressive enough to capture thecomplexity of the rules Status as ISO standard ensures stability and longevity Master topic Parent topic Child topic Guidance topic Derived topic Responsible person Concept Workflow state
IT asset management The University of Oslo is using the OKS to manage IT assets Servers, clusters, databases, etc are described in a TM This is used to answer questions like Service X is down, who do I call? If I take Y down, what else goes? If operating system Z is upgraded, what apps are affected? System driven by composite topic map Partly autogenerated Partly handcoded Two applications provide accessto the knowledge base Whitney: online Houson: offline (for use in emergencies) Houdini Whitney Syntax control OKS schema validation Versioning with CVS Navigator framework UIOTM FW OKS API OKS Engine RDBMS backend XTM usit.ltm(handcoded) oracle.ltm(generated) CVS
Asset management: Manufacturing The Y-12 plant at DoE is using the OKS to map its plant The purpose is to get an overview of equipment, processes, materials required, parts already built, etc.
Two main kinds Big application suites complete frameworks for building solutions engines at the core with end-user tools on top Smaller, open source tools many are just engines some are more specific tools for a single purpose
Ontopia Open source Java-based suite of tools engine + query engine generic ontology designer + instance editor conversion tools (RDBMS, RDF, XML, ...) presentation frameworks (JSP, portlets, ...) CMS integrations automatic classification graphical visualization web service interfaces browser ...
Web3 Commercial .NET-based suite engine + query engine Sharepoint integration built-in security model web service interfaces presentation framework
Topincs Web-based knowledge management tool wiki-like, but TMCL-based collaborative complex presentation features version 5.1 allows embedded programming in the TM
Wandora Open source Java-based application suite core engine presentation framework extensive set of input converters many export formats ontology designer + instance editor
topicWorks Commercial Java-based application suite core engine sophisticated data navigator Excel plugin ready-made ontologies
ZTM Open source Topic Maps-based CMS written in Python, on top of Zope used for a large number of portals (e.g vestforsk.no) very advanced CMS features enables very rapid development
TopicMapsLab SesameTM TMAPI implementation on top of Sesame triple store tmql4j TMQL query engine on top of TMAPI Aranuka object mapping library Onotoa graphical TMCL modelling tool Maiana social Topic Maps browser MajorTom virtual merging Topic Maps engine ...
Papers Topic Maps in Encyclopedia of Library Science http://www.ontopedia.net/pepper/papers/ELIS-TopicMaps.pdf The TAO of Topic Maps http://www.ontopia.net/topicmaps/materials/tao.html Metadata? Thesauri? Taxonomies? Topic Maps! http://www.ontopia.net/topicmaps/materials/tm-vs-thesauri.html