From Saltlux

Contents 1 Best Practices 1.1 Automatic Semantic Metadata Generation (Ontology Instantiation) 1.2 Semantic Search for Intelligent e-Marketplace 1.3 Semantic MDR and IR for National Archives 1.4 Personalized News Recommendation System 1.5 Ontology based Digital Library 1.6 Intelligent Global Patent Analysis System 1.7 Mobile Social Network Analysis System 1.8 Mobile Content Recommendation System 1.9 UI Auto-configuration for Personalized Mobile Service 1.10 Middleware platform for Context-aware Services 1.11 Intelligent Car Navigation for Ubiquitous City

Best Practices

Automatic Semantic Metadata Generation (Ontology Instantiation)

Background
In the last ten years, Korea has aggressively endorsed and invested into the development of government agency focused IT Intelligence Solution. So much so that it has earned the nickname, “IT Korea”.

As a result, Korea has since 2007 set forth a 5 year plan supported by the Ministry of Knowledge Economy to construct the core infrastructure essential to building a Korean knowledge based economy. To further this plan, under the supervision of KAIST (Korea Advanced Institute of Science and Technology), the following research organizations have been realized: Ubiquitous life, Ontology-based knowledge management System, IT knowledge portal, Ubiquitous market.

Ontology instantiation technology is key to building the core infrastructure. This technology semi-automatically extracts language that can be used as instances from the text or web document within the IT domain and by connecting to the National IT Ontology, which is used as a home base, is able to generate and track results.

The goal of the five year plan is to create an Instance production scale in the following chronological order: IT core Ontology, IT device Ontology, IT service Ontology, IT life Ontology. Therefore appropriate vocabulary relating to the above categories may be generated in the following order: 50 thousand, 20 thousand, 70 thousand, and 100 thousand.

Challenge
A systematic automated approach is necessary when appropriate vocabulary is extracted from the text or web document of an IT domain. After the vocabulary is extracted, the IT domain Ontology and the instance are connected to each other to generate the Instance. This process is manually conducted by an individual, this requires a large consumption of time and resources.

From the accumulated text or web document, extracting the vocabulary and connecting the instance to the Ontology needs to be a bundled automated process. However, at each step a manual examination needs to be undertaken so the entire process needs to be a part-automated process.

Solutions
The combination of Natural Language Processing, Named Entity Extraction and Ontology mapping technology is the solution to automatically generating over 1 million instances. This solution is named as the OPTIMA (Ontology Population Tool based n Information Extraction and Matching) system.

By applying the crawling application to the text or web document within the IT domain, all of these data sources can be
approached in the local files format or it can be downloaded as a file from the file server and can be used as input data for Named entity extraction.

Machine learning basis is named entity extraction which is CRF (Conditional Random Fields), ME (Mixtures of Experts), and SVM (Support Vector Machine) method.

The extracted named entity is recalculated into a usable ontology schema’s class and
Based on the property, a similarity calculation is conducted. Through this calculation, the mapping process is executed and the result is then stored in the repository in triple format.

In addition, human computing technology has been applied for product betterment. Numerous participants through a common workbench work together to edit and make product betterment decisions possible regarding Named entity’s validity and Ontology instance’s probability .



                     [System Architecture]


Benefits
The automation of the Ontology Instance generation allows for over 85% reduction in time and resource expenses. Large scale system of Ontology instance (semantic network) and use of reasoning engine’s use and domain specialist’s collaborative process allows for uniformed semantic data results and continuous product betterment.
 
Next Steps
By improving the capability of the Named Entity Extraction, relation (object property) analysis, which is more powerful than co-reference analysis technology, is currently under development. Additionally, by improving the standard algorithm the processing speed is accelerated to twice as fast as its current state.

The module Eclipse were plug-in based, presently these are also in the process of being converted to UIMA (Unstructured Information Management Architecture) platform based component.
Henceforth, various applications associated with UIMA may be easily utilized.

Client
Korean Government and Agency
Why Saltlux?
Saltlux is acknowledged as Asia’s leading semantically specialized corporation as it possesses the capability of natural language process, NE extraction and other basic technologies.
Background
Saltlux builds the core infrastructure such as ubiquitous life, ontology-based knowledge management system, IT knowledge portal, and ubiquitous market that are necessary for building Korean knowledge based economy.
Challenge
Extraneous expense and resources are necessary to manually generate the domain ontology’s instance from the text or web document of the IT domain. The challenge is to fervently develop the tool to automate the process of generating uniform quality results.
Solution
The combination of[IN2] Platform’s NLP based machine learning engine and semantic technology results in automatically extracting named entity from documents such as HTML, and text. This generates the construction of an automated ontology instance generation system.
Benefits
The automation of ontology instance generation process reduces time and expense by over 85%. The use of reasoning engine and the domain specialist’s collaborative work results in high quality large scale semantic network construction.
Next Steps
As the next step improvement of co-reference and relation analysis performance is being strengthened and UIMA environment will be supported.

Semantic Search for Intelligent e-Marketplace

Background
The global scale of market in the e-Marketplace is growing. The variety of goods and articles traded in the online marketplace are also increasing.

Korea’s online market scale increases yearly. According to the market trend reports, 4463 internet shopping malls are presently online. This is a 14% increase compared to 2006. The volume of trade of an internet shopping mall is at 13trillion and 4596million won, showing a significantly high yearly growth rate of 26.1%.In comparison to shopping malls, and discount wholesaler’s 8% annual growth rate, the internet shopping mall shows a much higher potential of growth.

In this internet marketplace, consumers are seeking to conveniently gain and exact accurate product information in a short period of time. In addition, consumers look for systems that can easily conduct product comparison analysis and purchase products through these systems.


Challenge
Majority of internet shopping malls utilize a product taxonomy navigation system where the keyword becomes the basis of the IR (information retrieval) method when searching for a product.

When the product taxonomy method is used to find a specific product, a search is undertaken and too many product catalogue results have to be searched through. If the user is not aware of the specification and option information, the user faces many complications in locating and purchasing the product.

When the product is located by IR, a basic keyword matching process is undertaken, thus there is a lack of accuracy and the user faces another challenge in comparing and getting specific product information.

In order to increase customer satisfaction in e-Marketplace product purchase, the user must be able to search product reviews and product related articles conveniently before purchase.


Solution
A semantic based IR system for the e-Marketplace automatically generates semantic meta-data and stores it in the triple repository. It then collaborates with the IR system that is based on query and reasoning technology thus producing an informative and accurate product information search.

The Semantic Meta-data system automatically extracts from accumulated product databases and the web. The extracted meta-data is then stored, in RDF format, in the triple repository together with the Ontology.

If the keyword based IR system produces a simple index searching capability, the developed semantic search system applies the triple repository to the query and the keyword index search as a hybrid. Additionally, by using the ontology, the synonyms and homonyms are treated efficiently. It is important to note that analogous product recommendation can be made possible through ontology subsumption reasoning, the concept expansion and narrowing.

The semi-automated semantic meta-data system takes in to account the detailed capabilities and options of the product so that the conclusive result is accurate product search and recommendation. Due to semantic meta-data’s rule based reasoning technology, the resulting information is of higher quality than those produced to clients of competitive analogous products.

In addition, this developed system goes through the mining process of blogs, news articles and accumulated web data related to the product to output product related review, evaluation and reports along with product specifications.


                    [System Architecture]

Benefits
After introducing an understanding based product information search system, it is able to produce superior quality information to the consumer by improving the accuracy and recall factors simultaneously regarding the information that the consumer is requesting. By not only providing the consumer with in-house product reviews, but also providing external product reviews (i.e. blogs, communities, etc…) and news articles, there has been a tremendous increase in consumer trust and satisfaction.

Additionally, by producing similar capability and specification product information, simultaneously providing alternative and related products and purchase information, results in increased customer satisfaction.
 
Next Steps
Henceforth, by increasing the reasoning skills, it is necessary to increase the ability of the system to find competitive products, accessories and related product information for the consumer.

It will be a reality of the near future to analyze consumer purchase patterns and product search habits to customize the product recommendation service function. An intelligent e-Marketplace system development will make this a reality.

Client
e-Commerce Service Providers
Why Saltlux?
Saltlux has provided information retrieval systems to over 300 customers. It holds a unique market leadership in the field of intelligent IR based on semantic technology.
Background
Consumers are not only provided the product information easily, quickly and accurately in the e-Marketplace, but also the customer needs for comparative analysis and product recommendation function are being increased.
Challenge
Most of the problems emerged during product search and purchase are based on the semantic ambiguity and insufficient meta-data. Unambiguous product information, secured meta-data and intellectual search, and consumer product evaluation must be provided all together to customers.
Solution
The intelligent e-Marketplace should be built through incorporation of DOR, the retrieval engine of [IN2]Platform, Optima, the ontology auto-creation tool and the SOR, the semantic framework.
Benefits
Accuracy and recall, two of the factors essential to product information search have both been improved. This will enable to recommend similar and competitive products, and the customer satisfaction will be improved through ontology and reasoning technology.
Next Steps
The next step is to provide individually customized service utilizing the consumer’s individual information such as user’s consumption pattern and the product search trend. An improved UI will also be provided to have user understood easily and instinct.

Semantic MDR and IR for National Archives

Background
NAK, the National Archives of Korea, is a Government service agency that accumulates and preserves historical records and data. It then makes this available through various means.

In January of 2004, the agency began an electronic organization of the records. A record system model of the central records management system was developed in 2006 and the technological analysis system and electronic document system were connected.

All of the records in the NAK and the National Government records system were interlinked and a new search engine was developed to search through both databases. It is the goal of the system to provide a one-day information delivery system to the public.

Challenge
Presently the NAK does not have a standard metadata storage organization. There is no job classification, department classification or contents type (text, books, periodicals, etc…). And depending on the usage objective, various heterogeneneous metadata maintenance organization is being done on a case by case basis. Because the NAK is associated with various sources, the semantic interoperability takes priority over concerns regarding its massive volume.

The NAK, because of the challenges referred to above, has recognized the need to develop a large capacity Semantic Information Retrieval System and construct Semantic Metadata Repository (SMDR) that will keep 70million non-current national records.
In order to build a collective search engine, there is a need to resolve the inhomogeneity issue between the metadata entity name, attribute and relation. In this project, the ISO/IEC 11179 MDR Metamodel needs to be executed to output over 10million RDF triple indexing and real-time query capacity.

Solution
The NAK’s search engine construction needs to be approached from a Top-down, ISO-11179 standard ontology MDR type. This is to confirm the interoperability and construct a domain ontology from a bottom-up approach. Thus over 12 million, large capacity metadata semantic IR System is constructed.

The core factor to the Semantic IR is Semantic annotation, this skill is essential to metadata search. Therefore we used a Text-Mining Solution [IN2]TMS to extract NAK’s relationships then took ontology based mapping solution, [IN2]SEMANO to execute the automated Semantic annotation.

Additionally, to search the large capacity, metadata, the [IN2]SOR framework and metadata’s storage AllegroGraph was utilized to realize a reasonable capability of the Semantic metadata.

The reasoning engine OntoBroker’ s reasoning rules are appropriate to produce semantic search results. This increases customer satisfaction dramatically.


                     [System Architecture]


Benefits
The previous keyword search system, directory system and collaborative search system took the keyword matching approach. These limitations were overcome by construction of NAK metadata relationship and reasoning based search engine service that resolves the user’s information approach and ambiguity.

Additionally, using the Web 2.0 technology, an improved user interface and collaborative information system search service to bring a reciprocity connection to the program was constructed.

The Text-Mining technology has improved the user’s experience in searching the appropriate information. Because the search execution is now approachable from an information and subject based type of search, even information that was previously difficult to understand has been simplified. Therefore there is an increase in the usage and the effectiveness of the system.

Next Steps
By participation in the EU and the FP7 LarKC project, the large capacity metadata reasoning technology is to be improved by the Semantic annotation technology. Through further development of Semantic technology based search engines, large capacity, metadata’s reciprocally related Semantic Information Retrieval Service is to be developed.

Client
National Archives of Korea (NAK)
Why Saltlux?
Saltlux’s semantic and search engine technology are necessary for building metadata repository and semantic search system.
Background
NAK manages all national records, and makes these records open for public access. They need to effectively integrate the distributed heterogeneous metadata system and to serve based on the system.
Challenge
The semantic annotation of non-current record of approximately 70 million cases and the triple indexing studies of over a billion cases are needed to be implemented. The semantic interoperability needs to be secured and the development of semantic search service system construction must be completed in a given period of time.
Solution
By utilizing Saltlux’s SOR and Semano of [IN2]Platform, ontology based MDR is constructed under the ISO-11179 standards to manage the cross operability of national records. The record information based domain ontology and a semantic search system of approximately 1.2 billion metadata cases have been completely constructed.
Benefits
Semantic oriented metadata incorporation and ontology and reasoning based search service will make improvement of user satisfaction of services.
Next Steps
Functional improvement of interoperability and reasoning of ultra-large scale metadata and growth and strengthening of the national service will be the next project to accomplish.

Personalized News Recommendation System

Background
The present day consumer wants to receive stock reports, news of competitors and news relating to themselves and their business as soon as it becomes available. This information becomes pertinent in the individuals lives or in the survival of a business. It is information that goes hand in hand to provide the competitive edge for businesses and social issues necessary in the daily lives of individuals.

Consumers hope to receive relevant news without obstruction. Due to technological advances and developments media distribution channels, news is not only received in print form but also distributed through the internet, cell phone and other terminals.

In addition, as globalization takes effect, the daily lives of individuals become increasingly complicated and users are able to publish online, thus the defined line between news and information blurs resulting in a large quantity of news type and volume.

Challenge
Standard PC users find pertinent news articles through internet news channel distribution services. These services allow the user to visit the site and read the article, or find relevant news articles through search engines. Therefore, it is through the user’s own accord that the users seeks news articles.

Following, when the user connects to a news distribution site, users can see the same interface as all other users or the user can select to customize the categories of news the individual wishes to browse through.

However, the user is placed in a variety of environments and it is the challenge and goal of news distribution services to figure out what is of most interest to the consumer in a customized news distribution setting.

Nevertheless, the user finds it difficult to self recognize one’s own interest this diverse information environment. A news distribution system that will resolve this entire information environment has many limitations.

Solution
The news recommendation system accumulates context information using the [IN2] Platform’s context crawler. This accumulated context is then appropriately extracted and distributed using the [IN2]TMS and the information is indexed and searched through [IN2]DOR. A multi-modal interaction technology is used for this customized search and recommendation service on various devices.

Accumulated user context information is as follows: present user location/time, user’s frequent information search category keyword, user schedule, weather information and other user profile information.

The above information is collected by a sensor Network and the individual’s profile is collected by the context crawler. All of this information is then combined through the news ontology and the reasoning engine is able to determine what news articles should be distributed through the customized news service to the individual.

As the news distributed through this customized service is extracted from a variety of news sources, only the text is finally output to the receiver thus the Wrapper module is used in this process.

Through the [IN2]SOR’s extraction technology, the appropriate news is selected for the user and depending on the user’s device the format of text delivery differs. For instance, on the PC the whole news article is distributed, but on the Mobile phone the news is automatically summarized by content.



                       [System Architecture]


Benefits
Users can, without the use of a PC, now search their interests and find recommendations through a mobile phone and various other portable terminals. According to the user’s preference, location, weather, economic situation and other various contexts, the recommendations are dynamically reconstructed and presented to the user in real time.

All of the news meta-data that are automatically extracted is saved and stored in a public storage RDF format. Thus this is a cost effective and flexible service system.

The multi-modality aspect of the news content is especially impressive to customers. Depending on the type of device customers can receive summary of the news in real time and the power of the sound output all increase customer satisfaction. 
 
Next Steps
It is foreseeable that this news recommendation service system will be utilized in various ubiquitous city projects in Korea. In this telematics service environment, wibro age, the multi-modal interaction capacity shows to be within the expanded business model of the IT service providers.

Client
Korea Telecom (KT)
Why Saltlux?
Saltlux already possesses the context awareness technology and reasoning technology that enables news crawling, automatic extraction and storage of metadata and experience.
Background
As the IT technology including mobile environment advances, consumer needs under the diversified context that individually customized information recommendation services are increased.
Challenge
By analyzing the individual’s preference and context, it was necessary to develop the technology to extract the information that is pertinent to the individual. Additional need was the development of content changing technology that matches to the device capability.
Solution
By using the DOR, TMS, SOR functions in the [IN2] platform, a system was developed to collect, select, restructure and recommend information for the users in accordance with the user’s environment.
Benefits
The users are now able to receive various customized information services in the mobile and telematics environments. Through this service, consumer feel much satisfaction and a new service model development is now in progress.
Next Steps
The multi-model interaction function will be improved and used in diverse u-city, telematics service environments.

Ontology based Digital Library

Background
The KERIS (Korea Education & Research Information Service) is Korea Government Agency’s largest scale online distributor of academic dissertation and research material. KERIS has introduced a variety of high technology to realize an advanced digital library, recently it has acknowledged semantic web technology as a revolutionary solution and is keenly examining its prospects.

In 2006, KERIS and Saltlux worked together for 10 months to construct an ontology based search system, related subject recommendation, intermission technology and service roadmap establishment project. 
 
Challenge
This project’s goal is to materialize semantic information retrieval system based on heavy ontology and reasoning technology. The first large scale domestic and international academic dissertation semantic IR system development began with the goal to open and make available its service to the public in real-time.

The largest challenge faced by this project was in the lack of academic metadata and reasoning engine’s performance.
 
Solution
In order to construct the Semantic IR System, the Ontostudio system, supported by F-logic, was utilized to construct 182 ontology classes, through a manual process 2924 high quality instances were generated. Additionally, through [IN2]Semano, 11thousand ontology instances were automatically extracted from the dissertations. The construction of the f-logic rule was more effective in extracting the instances regarding the various ontology and user’s query than the [IN2] SOR and OntoBroker.

The constructed ontology basis of reasoning and quality of the system is combined with that of the search engine [IN2]DOR and index level. To increase the Semantic IR system capacity, the Knowledge Base’s materialization and query chach technology was applied.


                                               [System Architecture]

Benefits
The Ontology based semantic information retrieval system has been proven to effectively resolve the problems that the previous keyword based search engine encountered.

The search of analogous concepts, and relevant concepts have been expanded and each subject centers around a navigation technology system with diverse class system that benefits researchers seeking information from the database.

In addition, the developed semantic IR system has proven to not only search subject and related author, related research institute and related professor but this rule based reasoning ontology system has shown the possibility of displaying the information and recommendation in a intellectual format.

This system is currently being tested at KERIS’s public service website, and is so far receiving positive feedback from its users.

Next Steps
The next steps are to expand on the limitations of the search domain and strengthen the relationship analysis search capacity to the research information found.

There will be a significant difference to be noted if the advanced semantic annotation’s instantiation technology is utilized and the improved capacity semantic repository system is upgraded.

Client
KERIS (Korea Education & Research Information Service)
Why Saltlux?
Saltlux possess the significant technology, products and the experience in building a Semantic Digital Library.
Background
KERIS was in need to advance its present online digital library system using the semantic web based service to build a service infrastructure.
Challenge
It was essential to build a large scale storage semantic IR system and a high performance reasoning system. The quality of the metadata of the original text was another challenge that needs to be overcome.
Solution
Using OntoStudio, the f-logic ontology and inference rule were designed and the semantic IR system was developed by collaborating the reasoning engine OntoBroker, the interrelated [IN2]SOR, and the search engine [IN2]DOR. [IN2]Semano was utilized for automatic generation of the large scale ontology instance.
Benefits
The developed ontology based semantic IR system provides expanded search results and generates concept relationship centered information navigation function. The diversified additional information recommendation function has dramatically improved the customer’s service satisfaction, and showed the possibility to effectively overcome the limitation of the current keyword based search system.
Next Steps
The next steps are to expand the scope of the search domain and to further improve the performance and functionality of the system.

Intelligent Global Patent Analysis System

Background
In analysis of trends in the technological market, the quantities analysis of the volume of patents is treated as a unique indicator. In recent years business models are acknowledged as individual patents and can be registered. Therefore, this patent analysis that was once only pertinent to technologically based sectors are now also significant in the service sectors.

The patent analysis not only affects technological development but has also expanded into business development decisions.

In order to conduct intellectual property rights analysis, abundant patent information needed to be processed and the appropriate application method evaluated. However at present, majority of intellectual property rights, and patent related analysis relies on the developer’s experience or keyword based search system. Therefore, there is a limitation to receiving quality information for analysis.

The LG Electronics patent center utilizes a text mining tool and automatically extracts information from unstructured patent text. They analyze the pattern, find similar patents and analyze the technology and research problems associated with it to find solutions to its technological problems

In addition, through the Ontology based significant keyword expansion, understanding based search and significant technology, accessories, system structure, technological query and other patent analysis systems were developed.

Challenge
An automated knowledge based precedent technology search system to search through all the applied and recorded patents was necessary. The demand for precedent technology similarity search, keyword and research comparison emerged.

Additionally, there was a need for precedent technology excavation ability that was hidden within the expanded keyword search. LG Electronics needed an in-house system to search all internal technology text documents and an automated system to search world patent database. This database search also needed to carry the function to classify in to different categories and view the text from various aspects. Therefore, precedent patent analysis system would give a fast and easy analysis of the technical query.

A solution needed to be found to analyze, research and confront the technology/patent problem, thus a conceptual and classified technological system to collaboratively resolve these needs emerged.

Solution
The [IN2]DOR and TMS allowed for the construction of the precedent patent analysis system. Keyword, text based knowledge precedent patent search also became possible as a result. Up to 2million documents could be searched through and the following languages of search could be used: Korean, Japanese and English.

Based on ontology, the keyword search was expanded and LG electronicsbusiness development/technology text/word clustering and large dimension automated analysis system, SVM model, was constructed. Superior quality of language automated patent analysis now became available in English and Japanese.

By constructing the automated community system, a superior quality Feature and Information extraction capabilities were realized. A template based automated summary generation system was also provided. In collaboration with the search system, a self-possessed search system became available.


                      [System Architecture]

Benefits
The patent analysis system was able to search recordation of application technology and confirm strategic ideas thus overcoming the limitations of the basic keyword search. Henceforward, the system was able to direct the direction of research and development. Additionally, it eliminated disputes in product and technology development’s intellectual property rights. It also contributed to patent acquisition and at time of patent application was able to confirm the technological extent and the rights of territory and expansion within the technology.

This system made possible individual professional field’s administration of intellectual property rights. The patent related research and analysis problems and solutions need to be automatically collected and outputted. The technology’s strengths and weaknesses can easily be identified. 
 
Next Steps
The next steps are to construct an intellectual patent analysis system by utilizing an even more powerful semantic technology. This development will allow for property’s increase in value and will transfer the intellectual property rights from a Cost center to being a Profit center. This development could play a key role in the development of new technology and products.

Client
LG Electronics
Why Saltlux?
Saltlux’s information search, mining and semantic technology already has a successful history of constructing patent analysis systems.
Background
As the value of patent increases, they are becoming the source of national technological economic power. Therefore, an advanced form of intellectual property rights analysis system development is necessary.
Challenge
An intellectual search system realization and multi-language international precedent technology patent system that takes care of automatic classification and clustering needs to be constructed based on ontology. 30 million studies and individualized search and analysis tools are all pressing goals to achieve.
Solution
By utilizing [IN2]DOR and TMS, the ontology based keyword expansion, word clustering, classification, automatic analysis of the SVM Model technology has been realized. Therefore the researchers can now realize and individualized service.
Benefits
Intellectual property rights have been improved, aggressive patent analysis has allowed for the strengthening of national economics and individual corporations. This advanced technology has allowed for a reduction in cost and increased satisfaction.
Next Steps
Even more powerful semantic technology than that in place is to be developed to increase value of intellectual property and increase the capital power of individual corporations.

Mobile Social Network Analysis System

Background
The key issue of the 2003 internet market was blogs. In 2005, 2006, the key work has become SNA (Social Network Analysis). SNA is a multifunctional system that searches within the internet, communities and services.

SNA has crossed the borders of from being a personal and community based social network on the internet to a mobile social network system. Within the telecom networks, the Social Network construction is mobile terminal user’s mapping with individual telecom network. With each nodes material and understanding relationship, uncover the meaning and provide an intellectual mobile service to the user.

There is still a large difficulty in effectively treating the large scale real-time Call detail records with Topology and Tie-Strength.

Recently in the Telecom market, due to the demands of the society, Korea Telecom Freetel(KTF) has applied the Mobile Social Network Analysis System to its corporation. By using this Semantic Technology, user’s can receive a variety of services.

Challenge
In the Mobile environment, the SNA needs to analyze and treat millions of large capacity communication information with the User’s Profile. Additionally, through the Data Mining basis, analysis of the information needs to be conclusive in finding a pattern.

Through the SNA, the Social Network’s centrality, betweeness, and closeness needs to be analyzed and each nod’s tie-strength also needs to be analyzed.

Solution
It is vital that this project’s large capacity call detail records are extracted fro relevant information. This information then needs to be placed in an entity relationship as ontology.

Thus, Saltlux has worked with numerous mobile corporations to develop a mobile social network analysis system based on the [IN2]SOR.

First, from the user information received from the mobile network, the first social relation was established. Using the call data, the second social relation was established according to location, call time and call method.

Data mining was able to examine the information collector’s accumulated information. Such information include, social information, personal relationship analysis module, social relationship analysis module, registered information and their personal family, friend and work relationships.

With the search results acting as the basis the social network editing process was edited. From the work done above, a Social Network Analysis System was constructed and many Social Network Services became available to the users.



                      [System Architecture]


Benefits
By using the social network in the mobile environment, the users can choose many cost effective methods. The users can also receive diverse intellectual mobile services.

Through the Social Network Analysis, the central Hub of the Social Network was discovered and become the center of various Mobile advertising and marketing projects. Users can receive contents recommendations, professional searches, share interests & Activities and even a higher quality of service.
 
Next Steps
As the mobile environment changes to a 3G base, regardless of location, users can receive a superior quality of service.

As well, the Mobile Social Network Service provides new information, exchange and real-life exchanges. This new environment and relationship generation foundation will be the basis of many new intellectual services to come.

Client
KTF (Korea Telecom Freetel)
Why Saltlux?
Saltlux possesses experience in providing world mobile telecom corporations with semantic based products.
Background
Virtual communities are emerging as an important social issue of which Telecom communities are being influenced. These portable mobile device users are requesting various intellectual services through the internet.
Challenge
The treatment of CDR Data and data to be analyzed, extracted and patterns deduced was the challenge. Additionally, the demand for the deduced Nodes and the tools to analyze them became increasingly sought after.
Solution
Information from the CDR Data was analyzed and this analysis became the basis of construction for telecom network ontology. Additionally social reference networks can analyze personal and social relationships, then the analysis can be searched through Data Mining.
Benefits
Common personal interests and activity groups can be formed and professional associations can be searched through this intellectual services can also be offered.
Next Steps
Social reference network formations will allow for diverse social relationship analysis where registered users will be able to use be provided with intellectual, customized services

Mobile Content Recommendation System

Background
The IP Multimedia Subsystem(IMS) and similar mobile telecom environments provide mobile telecom services on an IP base. Thus various and large capacity of contents and services are delivered to mobile telecom users.

Through this the mobile telecom users can receive seamless service regardless of whether it is wireless or not. Contents and service providers can expand on current business plans and launch new business plans. Mobile businesses can give additional value to its services.

However, this type of contents analysis can only figure out what types of contents and services that users are looking for. The development process has to look deeply in to what the user is actually requesting from these contents and services.

The consumers can however, receive many different contents and services an feels much satisfaction. The development and delivery process however has to be looked at with more detail and preparation.

Therefore the KTF(Korea Telecom Freetel) has realized the need for an intellectual Contents Recommendation System.

Challenge
To increase the user friendliness in terms of the contents and service, the correct delivery system has to be fitted for each user and the appropriate contents must be delivered to each user. At the appropriate time and appropriate approach much be take with each user.

However, the present mobile there are limitations in providing a collaborative organization system. Rather than this type of collaborative organization, the corporations are requesting for the development of intellectual services. The development of intellectual services are lacking in this type of environment.

In conclusion, to provide this type of customized contents and service, the information provided to the users
must be discriminately selected and provided to the user.

Solution
To discover the mobile telecommunication user’s mobile and LAN contents, the user’s registered information is categorized.

This discriminated information is sent through the contents-based filtering and collaborative filtering. Then this process is used to recommend the appropriate contents to the user.

The time, location, weather and schedule of the user along with the recommendable contents are also provided in a total contents recommendation pattern.

The user and the user’s social environment network is analyzed or the user can select a stereotype to input what type of customized contents they would like to have the service recommend for them.



                    [System Architecture]

Benefits
User, group, stereotype type of categories allow the user to select the contents to have recommended to them. The user’s property, knowledge and social environment analysis is conducted for context awareness based type of contents recommendation. The customized contents can reduce expenses, increase efficiency and customer can experience great satisfaction.

Additionally, because the service provider adjusts to the needs of the user, and provides intellectual contents, it increases its competitiveness and is able to not only maintain its customers but also increase its market share.
 
Next Steps
This project has proven success in the first year. This present Contents Recommendation Service’s use and the current program in place will provide various programs and various recommendation services to its users.

Client
KTF (Korea Telecom Freetel)
Why Saltlux?
Saltlux has gained situation recognition technology and recommendation technology through various projects, thus it is highly acknowledged for its Contents Recommendation System.
Background
Telecom corporations next generation mobile telecom attempts to send enormous volume of contents without any hinges in a customized format to its subscribers. Telecom corporations wish to reach a larger audience and create more business opportunities for itself.
Challenge
For the betterment of the user’ s approach to contents and service, the user’s personal interest and contents are analyzed. Through this analysis the delivery technology, of the Right Contents, Right Time and Right Way, is being sought after.
Solution
The [IN2] Recommendation Module materialize as a solution to the challenge. This module’s basis is in a Contents-Based Filtering, Collaborative Filtering system.
Benefits
In a mobile environment the appropriate contents can easily be uncovered and the information approach expenses can be minimized. Therefore the user finds the Contents usage to be friendly and convenient.
Next Steps
It is in the next steps to collaborate with the system in place to create a diverse Intelligent Recommendation Platform.

UI Auto-configuration for Personalized Mobile Service

Background
The presently distributed CDMA, WCDMA, WLAN and LAN based xDSL, FTTH, in the next generation All-IP to be totally integrated . This is to provide a diverse and large-volume of contents and services to the users.

The mobile communication user’s contents, service, cell phones, PDA, PDP, UMPC full screen mobile terminals are using a convergence mobile terminals. These type of devices have much smaller screens and smaller capacity compared to PCs. These also have inconvenient input systems.

Mobile telecommunication providers are constantly giving tireless effort to providing customized efforts so that the information delivery can be user friendly.

Following, KTF and other telecom providers are finding the increase need to deliver information in a user-friendly manner through various mobile terminals.
Therefore they are investing to developing their technology.

Challenge
In the next generation field, various mobile terminals users want to receive the same contents and service regardless of the type of device they are using. Contents and Service providers however, must deliver the information keeping each device’s property capability in mind.

Additionally, various mobile terminals find difficulty dealing with input manner and output screen. The user finds it extremely difficult to obtain the information that the individual is looking for.

Solution
The Mobile terminal’s disadvantages are recovered by distributing the information in an UI XML format. This way information can be appropriately formatted in whichever device the information is being outputted.

This type of expression in the National Mobile Standard Association’s Open Mobile Alliance (OMA)’s Dynamic Contents Delivery (DCD) is the national standard. The OMA, is sending contents in real time mobile device delivery through LAN lines.

In this project the devices are expressed in RDF’s CC/PP format. Through [IN2]SOR it is being maintained. The user profile, preference, ontology is saved in semantic metadata format. User context based reasoning rules used as the basis to redeveloping device UI.

In conclusion, Customizing UI needs to be generated to construct a personalized UI so the services can be selected according to personal preferences. Then the contents presented in the UI will be personalized contents according to preferences selected.

In delivering the individual preferences based contents recommendation, the pop-up format is used. This standard allows customized content delivery to individuals in real time.


                          [System Architecture]

Benefits
Telecommunication service providers are able to be independent from the limitation of various mobile devices. They can deliver contents in a personalized format. This new generation method will provide a content with a total interact and environment construction.

Customer service satisfaction will be greatly increased as terminal users will regardless of the service, construct and customize the information for distribution to its customers.
 
Next Steps
The above project has completed the stage one process of UI administration capability and UI construction information protocol standardization. Capability expansion possibilities for this prototype is now possible with this new development.

Henceforward, this system’s dynamic UI generation technology will be commercialized and diverse Mobile Device appliances will be expanded.

Client
KTF (Korea Telecom Freetel)
Why Saltlux?
Saltlux holds unique experiences and has found solutions in applying semantic technology to mobile and LAN Telecom corporations through a variety of projects.
Background
Regardless of the type of terminal that the user is using, telecom corporations are looking for technology and system construction that will customize the service and contents. These aim to better customer satisfaction.
Challenge
The input method, visual display and various character price all factor to create challenges in the user’s interaction with the service and contents.
Solution
The mobile device’s profile and user profile and preferences are stored in semantic metadata format in the [IN2]SOR. Through reasoning, the user customized UI auto-configuration process takes place. Through this a variety of formats for the mobile device, user customized content delivery and service recommendation are all possible.
Benefits
Regardless of the type of user’s terminal, the user can receive the desired service and contents easily thus customer satisfaction has increased. Additionally, this has made possible for telecom corporation’s next generation mobile communication network to constructed a new interface and expand its services.
Next Steps
This dynamic UI generation technology has created many possibilities for dealing with and expanding the different terminal interactions.

Middleware platform for Context-aware Services

Background
Intellectual type Service Middleware is the basis for Smart Home. Home Automation System is essential for the future of our cities. For a long time, enormous interest has been shown in this field and we have already realized many of these potentials as realities.

Ubiquitous technology takes simple formats of ordinary applications and with intelligent computing, they bring forth changes to the environment rather than physically applying the changes. To construct an intelligent environment, each system and service’s diverse device, mobile device needs to change quickly.

Intellectual Service Middleware goes beyond the simple format, high-speed internet and public environment equipment. It also applies to home network technology, information home appliance technology, software platform solution and contents’ diverse IT areas. A Context Awareness module needs to be developed for a new generation service environment.

Additionally, the digital home, intelligent Robot, Telematics Service must be combined to construct this Ubiquitous Society. To adapt to various environments and be able to indicate that, the most appropriate adjustments to the context fitting service environment will be possible.

Challenge
Context Awareness Service will take the user’s preferences and Environmental Context Information and adjust to the movement of the user. It has to be a customized service appropriate for each individual before distribution.

User centered intelligent service can be administered by solving the semantic ambiguity of the context information. A valid service must be found to fit the context and dynamically reconfigure to the Context Awareness Service Middleware. This is the challenge to be overcome and one that the user needs to find an solution to.

This middleware development must solve the conflict that arises from the sensor network’s heterogeneous context information and the interoperability. Additionally, for real-time service, large capacity context information storage, administration technology and service, discovery and adaptation skills are necessary.

Solution
Saltlux has developed a service middleware, [IN2]S-Cube. It takes large capacity of context information, accumulates, changes and saves for public administration.

This developed middleware takes users interest and proactively customizes to the context. With the information context, physical and environmental context combination changes the administration to semantic metadata. The Context metadata’s expression, is administered and reasoned through OWL DL based ubiquitous ontology. The appropriate service according to the user context is found, and for adaptation, the f-logic based reasoning engine, Ontobroker, and SOR are processed together.

Various ubiquitous environment and devices is using the OSGiframework as its standard format. Multi-modality has also been constructed in the UI device customization through the W3C’s EMMA.


                        [System Architecture]


Benefits
This solution can be applied to intelligent Service Middleware, home network service, Telematics, intelligent robot departments.

By using the developed middleware in the home network, each user can customize the information they want to receive and receive such information. Additionally, in the Telematics area, while in a car and driving, one can verbally voice requests to receive traffic information, schedule information and webmails.

This developed context-aware service middleware has lowered costs and has also lowered the development period by over 30%.
 
Next Steps
This developed service middleware technology will be used not only in the home network, telematics but also as intelligent robot, IPTV, Real-time Enterprise environments, u-city, Wibro next generation internet service areas.

Client
KT (Korea Telecom)
Why Saltlux?
Saltlux has sufficient experience in development Context representation and repository. It possesses the resources to apply ontology and reasoning technology.
Background
Context-aware service is a futuristic service model that is to be used in an ubiquitous computing environment. To be cost effective and market applicability a common middleware development became essential.
Challenge
The challenge was to remove large scale context information’s semantic ambiguity, confirm interoperability, and develop dynamic service reconfiguration and service adaptation technology.
Solution
Saltlux’s [IN2]S-Cube is a combination of ubiquitous ontology based semantic context metadata, offering interoperability that provides a uniform results system and the [IN2]SOR . The [IN2]SOR is interlined as a context-aware service adaptation technology. The S-Cube uses the OSGi and EMMA standard.
Benefits
In various ubiquitous environments, this middleware technology has lowered the cost of intelligent service development, and decrease the development time by over 30%.
Next Steps
The developed middleware will be used as part of the infrastructure in a variety of service environment, especially the IPTV and u-city as part of the infrastructure.

Intelligent Car Navigation for Ubiquitous City

Background
Globally, the new intelligent real-time city structure project is now in progress. In Korea, the project takes on the name ubiquitous city and is currently in progress.

‘u-City’ is a ubiquitous computing based environment, thus what is physical is intellectualized. Roads, cars, buildings and homes that physically exist plus all things that take up electronic space, cell phones, PDA, DMB devices are all interlinked and a new 3rd city space is created. This is the goal of this project.

The development of the infrastructure that can be anytime, anywhere, any information, any device without any obstruction in a seamless connection is the service reality of the u-City.

Challenge
Intelligent Transportation System(ITS) uses revolutionary technology to provide a new type of transportation service to increase the satisfaction of the urban inhibitors.

In the u-City environment, ITS plays a very important role. A convenient transportation system and traffic allows for increase efficiency in the whole city, Additionally it decreases the city operation fees. An open traffic service increases the satisfaction of the individuals who reside there.

ITS takes the accumulated data from the sensors locate on many of the city roads and analyses them in real time. For this the traffic context real time, administrating, filtering and technology development is necessary.

The car navigation system on the road cannot demonstrate a real time effect. Many users of the navigation system complain because it cannot anticipate road conditions. To resolve this complaint and to resolve heavy city traffic jams, an interlinked ITS car navigation system needs to be intellectualized. The Intelligent car navigation system will change course by itself, and recommend different road courses to take.

Solution
This project takes the [IN2]SOR and stores the traffic information as semantic metadata. Then through ontology reasoning traffic context awareness technology is constructed.

Saltlux has developed the f-logic rule for ITS development. It expresses all situations of the car, road and traffic conditions as an ontology relationship.

The result is the development of the intelligent car navigation pilot system It has proven to be effective in various traffic conditions and has expanded the possibilities possible through the navigation service.

This newly developed navigation system takes the departure and destination points and in real time, real situation, suggests to the driver alternative routes. This ontology based reasoning system is an extremely effective navigation method.

The significant technological challenges encountered during the development were scalability and reasoning performance. This recently developed system can in 1day million of real time scenarios can be developed in about 5 seconds.


                      [System Architecture]


Benefits
Intelligent Transportation System uses a variety of traffic sensors and car sensors to help with safe driving. This intelligent service offers many conveniences to the user. Recently developing u-city projects differentiate themselves through this type of service.
 
Next Steps
This intelligent system’s contextual awareness engine’s use produces various vertical market solutions. These solutions will play an important role in launching new values systems. Ubiquitous city’s total administration center and relationships are being pursued. The ITS standard platform format will be utilized for advancing many more projects.


Client
System Integrators for u-city projects
Why Saltlux?
Saltlux has the development experience and the products such as the context aware service system that is based on semantic technology. It also has experience in u-city development experience.
Background
The u-city vision is to build a 3rd city space that combines the physical and electronic space into one. With ITS and groundbreaking infrastructure this city can be constructed.
Challenge
The future ITS, large scale transportation sensor information ,must be able to collect real-time information, save and maintain it. It must but be able to automatically adjust to various situations. Intelligent car navigation and practical service models need to be expanded and tested for usability.
Solution
The ITS holding knowledge information ontology was constructed. Additionally by using the [IN2]SOR, the traffic information was converted to semantic metadata, stored and the result was the development of the intelligent car navigation service system.
Benefits
The ITS includes all traffic situations and provides superior customer support service and allows the user of the device with safe traffic service.
Next Steps
The next step is to standardize this service as a standard platform in Korea’s various u-City developments.