=Paper=
{{Paper
|id=None
|storemode=property
|title=Computing Recommendations for Long Term Data Accessibility basing on Open Knowledge and Linked Data
|pdfUrl=https://ceur-ws.org/Vol-811/paper8.pdf
|volume=Vol-811
}}
==Computing Recommendations for Long Term Data Accessibility basing on Open Knowledge and Linked Data==
https://ceur-ws.org/Vol-811/paper8.pdf
Computing Recommendations for Long Term Data
Accessibility basing on Open Knowledge and Linked Data
Sergiu Gordea Andrew Lindley Roman Graf
AIT - Austrian Institute of AIT - Austrian Institute of AIT - Austrian Institute of
Technology GmbH Technology GmbH Technology GmbH
Donau-City-Strasse 1 Donau-City-Strasse 1 Donau-City-Strasse 1
Vienna, Austria Vienna, Austria Vienna, Austria
sergiu.gordea@ait.ac.at andrew.lindley@ait.ac.at roman.graf@ait.ac.at
ABSTRACT 1. INTRODUCTION
Knowledge based recommender systems (KBRs) as natu-
Digital access to our cultural heritage assets was facilitated
ral followers of expert systems are nowadays used for sup-
through the rapid development of the digitization process
porting the decision making process in multiple application
and online publishing initiatives as Europeana or the Google
areas as: e-commerce, financial services, tourism, etc. One
books project. As Galleries, Libraries, Archiving institu-
of the most important challenges of KBRs is the construc-
tions and Museums (GLAM) created digital representations
tion of their underlying knowledge base. This is typically
of their masterpieces new concerns arise regarding the long-
composed by sets of factual knowledge, i.e. information de-
term accessibility of digitized and digitally born content.
scribing the application’s domain and business rules. Both
Repository managers of institutions need to take well docu-
together enable the drawing of conclusions and support the
mented decisions with regard to which digital object repre-
decisions making process when analyzing the utility of a spe-
sentations to use for archiving or long term access to their
cific item in a given context as for example, analyzing the ef-
valuable collections. The digital preservation recommender
fectiveness of digitizing and publishing Mircea Eliade’s book
system presented within this paper aims at reducing the
”History of Religious Ideas” within Google books.
complexity in the process of decision making by providing
Even though the world wide web has turned out to be
support for classification and the preservation risk analy-
the largest knowledge base, information published lacks an
sis of digital objects. Technical information which is avail-
unified well-formed representation and mainly is intended
able as linked data in open knowledge sources facilitates
for human readers. The Linked Open Data (LOD)1 and
the construction of the DiPRec’s recommender knowledge
Open Knowledge2 initiatives address these weaknesses by
base. This paper presents the DiPRec recommender sys-
describing a method on how to provide structured data in a
tem, a community approach on how to achieve the genera-
well-defined and queriable format. By linking together and
tion of well founded and trusted recommendations through
inferring properties of di↵erent independent and publically
open linked data and inferred knowledge in the domain of
available information sources like FreeBase3 , DbPedia4 and
long-term information preservation for GLAM institutions.
Pronom 5 within the specific context of a digital preservation
scenario we shortcut the well known challenge of KBRs, the
Categories and Subject Descriptors knowledge acquisition bottleneck.
H.3.7 [Information Systems Applications]: Digital Li- In this paper we present our work carried out in the con-
braries; M.8 [Knowledge Management]: Knowledge Reuse text of the Assets6 project with the aim of preparing the
ground for digital preservation within Europeana7 . The Eu-
General Terms ropeana portal serves as a central point for the large public
to easily explore and research European cultural and sci-
Digital preservation, Recommender systems entific heritage online. It aggregates and collects data on
digital resources from galleries, libraries, archives and muse-
Keywords ums accross Europe and by now manages about 19 million
object descriptions collected from more than 15 hundred in-
Knowledge based recommender, open recommendations, linked
stitutions. Within this very heterogeneous context it is eas-
open data, preservation planning
ily understandable that digital objects are encoded in very
heterogeneous file formats and versions throughout various
di↵erent hardware and software content repository systems.
Depending on the underlying use case it is likely that mul-
1
http://linkeddata.org/
2
http://www.okfn.org/
3
http://www.freebase.com
4
http://dbpedia.org/
5
http://www.nationalarchives.gov.uk/PRONOM/
6
http://www.assets4europeana.eu/
7
http://www.europeana.eu/portal/
51
tiple representations of the same ’physical’ object exist at explained in detail through a concrete example on the TIFF
a time. For example in most cases it is useful to provide file format. The evaluation of our approach is presented in
access copies on demand which are easily distribuatable via Section 4 by analyzing the digital collections of the Assets
the web while the master record needs to adhere to di↵er- project. This is followed in the last Section of the paper (nr.
ent requirements as for example the institution’s long-term 5) by the summarization of the concluding remarks for our
scenario and preservation policy. work.
A key topic in preservation planning is the file formats
used for encoding the digital information. The Pronom
Unique Identifiers (PUIDs) registry provides persistent, unique
2. RELATED WORK
and unambiguous identifiers for file formats and therefore Knowledge Based Recommender systems gained broad pop-
takes a fundamental role in the process of managing elec- ularity in e-commerce and e-tourism [7, 11, 24, 19] appli-
tronic records. Currently it lists information on about 820 cations supporting customers in their decision making pro-
di↵erent PUIDs. While some of the formats are properly cesses. The two most popular use cases are guidance through
documented, open-source and well supported, others may large and complex product o↵ers (e.g. trip organization, fea-
be outdated, redeemed by software vendors and no longer ture selection of technical equipment) as well as accompany-
functional in modern operating systems. As always the the ing the process of high cost decision making (e.g. financial
binary file’s dependencies on the underlying platform, its investments). When designing the DiPRec recommender we
configuration (codecs, plugins, etc.) as well as the render- took into consideration the Advisor Suite [12] and Planets
ing software are responsible on generating a concrete user Testbed infrastructure [16]. The main component of the
performance, it is vital to have a solid understanding on all Advisor Suite is a multipurpose workbench which o↵ers sup-
of them. This process is costly and requires a high degree port and advanced graphical user interfaces for constructing
of engineering expertise. Many of the GLAM institutions knowledge based recommenders. Advisor Suite features in-
already outsource IT related activities and don’t have the clude the import of product catalogues, visual editing of a
resources to keep track of the required level of complexity in recommendation workflow and the generation of a runtime
house. environment. The Planets8 project focused on constructing
The Digital Preservation Recommender (DiPRec) system practical services and tools for establishing empirical evi-
addresses the topics of ’preservation watch’ and ’preserva- dence in the process of informed decision making in the area
tion policy recommendation’. It proposes a solution in the of digital long-term preservation. A major achievement was
domain of digital long-term preservation for making doc- the definition of basic nouns and verbs for core preserva-
umented recommendations based on risk scores, while the tion operations. This allows to easily combine and swap
underlying knowledge base is built through a linked data ap- tools within a preservation workflow and lead to a num-
proach. Information from FreeBase, DbPedia and Pronom ber of over fifty preservation services. Available services
in the areas of file formats, file conversions tools, hardware were deployed and tested within the Planets Testbed [22], a
and software vendors is taken into account. The main con- uniform environment for experimentation under well-defined
tribution of this paper consists in the integration of open and controlled surroundings. It provides automated quality
(general or domain specific) data when constructing knowl- assurance support for tools like DROID9 , JHOVE10 and the
edge based recommendations. The ”knowledge acquisition eXtensible Characterisation Languages11 [5].
bottleneck” and the high costs of setting up and maintain- A key topic in preservation planning is the process of eval-
ing KBRs are still an impediment for extensively adoption uating objectives under the limitation of well-known con-
by the industry. Recommendations provided by DiPRec are straints. A state of the art report on technical require-
meant to support GLAM institutions across Europe in the ments and standards as well as available tools to support
process of analyzing their digital assets. The technical foun- the analysis and planning of preservation actions is given in
dation and the explanation of the DiPRec recommendations [2]. Strodl et al. present the Planets preservation planning
are computed on top of shared and collaboratively built data methodology Plato12 by an empirical evaluation of image
sources, trust in the area of LOD and digital preservation scenarios [21] and demonstrate specific cases of recommen-
is a key issue which has been left out for this paper due to dations for image content in four major National Libraries in
simplicity. Europe[4]. After eliciting information regarding the preser-
The novelty of our work consists in combining expert tools vation scenario (user requirements) the Plato tool is able to
(as File, Droid or Fido) and automated object identifica- recommend specific preservation actions [3] for a given sce-
tion processes, with structured information (e.g. techni- nario. The tool was specifically designed to work on sam-
cal information on file formats) from open data reposito- ples of the underlying data set and therefore is able to make
ries. This information is use for infering new knowledge, use of XCL or similar tools for automated quality assurance
calculate preservation risks and finally for computing rec- and semi-automated evaluation of objectives. In contrast to
ommendations on preservation actions in the domain of dig- these scenario evaluations, DiPRec aims at collecting infor-
ital long-term preservation. We present the rationale used mation on a broader range from open linked data registries
for the construction of the DiPRec recommender by pre- and dynamic knowledge sources. It can evaluate more gen-
senting concrete examples of a given content analysis which eral, even ’non-technical’ objectives (e.g. what is the risk
was provided for the Assets project. The rest of the pa- that no software vendor will support old formats like Word
per is organized as follows; in Section 2 we present related 8
http://www.planets-project.eu/
work carried out on recommender systems and in the field 9
http://droid.sourceforge.net/
of digital preservation. Section 3 highlights the architecture 10
http://hul.harvard.edu/jhove/
of DiPRec by comparing it against the construction of clas- 11
http://planetarium.hki.uni-koeln.de/public/XCL/
sical KBRs. The functionality provided by our system is 12
http://www.ifs.tuwien.ac.at/dp/plato/intro.html
52
Open Domain
Knowledge
Domain Expert Pronom UDFR
Knowledge Acquisition
Domain Expert Domain Knowledge
User requirements User Requirements
Knowledge Aggregation
Knowledge Base Recommendation Engine Knowledge Base Recommendation Engine
Domain Information Recommentations
Domain Information Recommentations Aggregation
Acquisition
(description and data)
Format
Identification DbPedia FreeBase
Linked Open Data
a) Classic KBR recommender b) DiPRec recommender
Figure 1: A comparison of regular KBRs and DiPRec recommender processes
3 documents? ). This is a significant improvement over the tion than the domain specific KBRs. Within the DiPRec
Plato tool where all this information needs to provided by recommender the Domain Information Aggregation module
domain experts. is responsible for collecting file format related information
The Scape13 project is one of the major current initiatives (e.g. formats, vendors, applications, etc.) from the open
[18] which is partially funded by the European Union’s FP7 knowledge bases Pronom, DBPedia and Freebase. Further-
on institutional preservation requirements. The project ad- more the Domain Knowledge Aggregation module combines
dresses besides the issues of scalable preservation and quality- the outcome of a risk analysis process with the knowledge
assured preservation workflows also the topic of policy-based manually provided by domain experts. Figure 1 compares
preservation planning and watch. the process used by regular KBRs and the one presented by
The paradigms of semantic Web and linked open data [6] DiPRec which enhances the process of building the under-
transform the web from a pool of information into a valu- lying knowledge base. In the following sections we present
able knowledge source of data according to the definitions extended details on how the knowledge base of DiPRrec is
of a knowledge management theory [17]. The exploitation built by using as example the Tagged Image File Format
of linked data as knowledge source for recommender sys- (TIFF).
tem started as research topic in the last few years and was The TIFF format is still very popular among the publish-
first applied to improve case-based and collaborative filter- ing industry, as it is a very adaptable file format although
ing recommenders [10, 9, 20]. In [20] the authors present the it did not have a major update since 1992. It was originally
Talis Aspire system which is able to assists educational sta↵ created by Aldus and since 2009 it is now under control of
in picking educational web resources. The employment of Adobe Systems. There are a number of extensions avail-
linked data in collaborative filtering and case-based reason- able (e.g. TIFF/IT, TIFF-FX) which have been based on
ing was explored by Heitmann and Hayes in [9] and [10]. the TIFF 6.0 specification, but not all of them are broadly
used. A standard and broadly accepted approach in the
3. SYSTEM OVERVIEW archiving world is the migration of TIFF encoded content
to the JPEG2000 format. In [4, 2] one can find the context
Typically the creation of classic knowledge based recom- in which several content providers took the decision to per-
mender systems consists of three main tasks. Dealing with form this kind of content migration. However within these
the collection of detailed descriptions of products o↵ers is scenarios, the context evaluation and the recommendation
followed by the process of constructing a recommendation were computed by domain experts and by expert systems.
knowledge base (see section 3.2). At runtime user require- The DiPRec system, on the one hand applies to the ap-
ments elicitation takes place and recommendations are com- proach of well-documented and trackable decision making,
puted based on the underlying recommendation knowledge and at the same time it uses a semi-automatic approach
base and the items that match the given user requirements. on domain knowledge acquisition. This reduces the human
DiPRec follows the same process but improves the way the e↵ort invested by domain experts when providing reserva-
knowledge base is built in order to reduce the e↵orts spent tion recommendations, reduces the financial e↵orts invested
on domain knowledge acquisition. This is especially relevant in the context evaluations, and in the same time is able to
for being exploited in GLAM preservation scenarios, where o↵er good quality recommendations.
the underlying knowledge base contains broader informa-
13
http://www.scape-project.eu/
53
FILE FORMAT DESCRIPTION
Format Name Tagged Image File Format (P), Tagged Image File Format (D), Tagged Image File Format(F)
Pronom Id fmt/10 (P)
Mime Type /media type/image/ti↵-fx, /media type/image/ti↵ (F), image/ti↵(P)
File Extensions .ti↵, .tif (D)
Current Version 6 (P)
Current Version Release Date 03 Jun 1992 (P)
Software License Proprietary software (D)
Software QuickView Plus, Acrobat, AutoCAD, CorelDraw, Freemaker, GoLive, Illustrator, Photoshop,
Powerpoint (P), SimpleText, Seashore, Imagine (D)
Software Homepage http://adobe.com/photoshop(D)
Operating System PC, Mac OS X, Microsoft Windows (D)
Genre Image (Raster) (P), Image file format (I), SimpleText - Text editor, Adobe Photoshop - Raster
graphics editor (D)
Open Format none (P)
Standards ISO 12639:2004 (W)
Vendors Aldus, Adobe Systems, Apple Computer, now Apple Inc., Microsoft (D), Adobe Systems
Incorporated (P), Aldus Corporation (P)
VENDOR DESCRIPTION
Organization Name Adobe Systems
Country United States (P)
Foundation date Dec 1982 (F)
Number of Employees 6068 (Jan 2007), 8660 (2009)(F), 9,117 (2010)(W)
Revenue 3,579,890,000 US$ (Nov 28, 2008) (F)
Homepage http://adobe.com/photoshop(F)
Table 1: File format and vendor description. (Information sources P = Pronom, D = DBPedia, F = Freebase,
W = Wikipedia)
3.1 Domain Information Aggregation considered to fulfill this requirement by combining the
Di↵erently to the e-commerce domain where KBRs import properties like ”NUMBER OF EMPLOYEES”, ”VEN-
detailed item descriptions from product catalogs there is no DOR REVENUE”). See Table 2.
such catalog for computer file formats. The Unified Digi- When aggregating domain information we are interrogat-
tal Format Registry (UDFR)14 project was started in 2009 ing external knowledge sources like DbPedia and Freebase
by a group of Universities and GLAM institutions with the which manage huge amounts of linked open data triples.
aim of building a single, shared technical registry for file This allows us to extract fragmental descriptions on file for-
formats based on a semantic web and linked data approach. mats, software applications and vendors supporting given
The project is based on the Pronom database which pro- file formats (see Table 1). DbPedia allows to post sophis-
vides basic information about a large number of file formats ticated queries using SPARQL query and OWL ontology
and will be extended by data on migration pathways and languages [13] for retrieving data available in Wikipedia.
available software/tools. The registry should be available Freebase [15] is a practical, scalable semantic database for
from the beginning of 2012. As Pronom data is not rich structured knowledge and is mainly composed and main-
enough to build a recommendation and reasoning mecha- tained by community members. Public read/write access to
nism for preservation scenarios of file formats on top, we Freebase is allowed through an graph-based query API using
collect additional information sources and aggregate them the Metaweb Query Language (MQL) [6]. PRONOM data
into a single homogeneous property representation in the is released as linked open data and is accessible through a
recommender’s knowledge base. DiPRec uses two types of public SPARQL endpoint.
operations for aggregating domain information:
• data unification: the data representation retrieved from AGGREGATED PROPERTIES
di↵erent knowledge bases is unified and combined un- File format related
der the DiPRecs property model definition. For exam- Is supported by major software vendors? yes
ple, the number of software tools supporting a given file Is an open file format? no
format is calculated over di↵erent data sources. The Is widely supported by current web browsers? yes
Which versions officially supported by vendor? 6.0
individual object’s namespace, the transformation pro- Which versions are frequently used? 6.0
cess of values, the query on how to extract a given Image file compression supported? yes
record, etc. are preserved and are part of the prop- Preservation related metadata
erty’s model representation. Is creator information available? yes/no
Is publisher information available? yes/no
• property composition: more abstract properties which Is digital rights information available? yes/no
require a hierarchical composition are computed by ag- Is file migration allowed? yes/no
gregating basic properties by weighted numbers. The Object creation date? datetime
Is an object preview available? URL
model on property definition is meant to be kept very
simple. For example ”supported by major vendors”
Table 2: Sample compound properties.
will check if at least one of the software companies is
14
http://www.udfr.org/
54
Media File Domain Knowledge
Aggregation
PRESERVATION
RISK
property set and consequently contributes to the risk com-
Score / Report putation over a given dimension is modeled through the in-
troduction of specific weighting factors (see Equation 1).
The value of the overall risk score for a given collection
Property Sets
of objects is computed as a weighted sum over all digital
preservation dimensions:
Domain Information X X
Aggregation Ri = wps,i ⇤ wp,ps ⇤ d(p, P F V (p)) (1)
ps2P Si p2P ROPps
DIGITAL
PRESERVATION
TOOLS Open Knowledge
(Droid/Pronom) FILE METADATA Sources Where Ri represents the preservation risk computed over
(DBPedia, Freebase,
PRONOM ) the dimension i. ps represents the index of the current prop-
erty set within all sets associated to the dimension i. The
Figure 2: Domain knowledge aggregation process. w(ps,i) is the weight of the contribution of the property set ps
3.2 Domain Knowledge Aggregation to the dimension i. Similarly, p stands for the index of cur-
rent properties within the list of properties available in the
Pronom as presented before is a viable resource for any-
given property set P ROPps . wp,ps denotes the importance
one requiring impartial and definitive information about the
of a property p for the property set ps. The distance be-
file formats, software products and other related data. Ex-
tween the current property and the defined - ’preservation
tremely valuable to the DiPRec recommender is the infor-
conform’ - value for this property is represented through
mation related to the file conversion tools based on a given
d(p, P F V (p)).
PUID. Therefore we employ the Droid 15 characterization
service for automatically extracting technical metadata and
identifying file formats from physical media files. This meta- 3.3 User requirements elicitation
data is then used in conjunction within the domain knowl- DiPRec is designed to work as a multi-purpose digital
edge aggregation process presented in the Fig. 2 preservation support tool which can be used in various sce-
The risk analysis module is in charge of evaluating in- narios by di↵erent types of customers. For examples the tool
formation previously aggregated in the DiPRec knowledge may support content providers in analyzing the ’preservation
base for a given record at hand over following (exemplary) friendliness’ of their infrastructure, their archiving solutions
dimensions of digital preservation: or the visibility of their artifacts published in the Europeana
portal. Recommendations are always to be seen in the con-
• Web accessibility: Dissemination copies are published text in which the digital objects are used. Within the scope
and accessible on e.g. the content provider’s web por- of the Assets project there is the common interest to o↵er
tal. There should be previews of objects (e.g. thumb- public access to digital assets through the Europeana portal
nails for images, video summaries, short intro for au- (i.e. web discovery), to provide advance search functional-
dio files) and ’rich’ object descriptions to increase their ity (i.e. description richness and preservation of provenance
visibility and retrieval. The chosen file representations information) as well as the topic of the data archiving di-
should render in the latest browsers without plugin mension.
support and cope with modern features (e.g. pseudo As a result of the requirements elicitation process user pro-
streaming, progressive image display, HTML5, X3D, files are created. A set of multiple choice questions is used to
etc). Content is made available through di↵erent ex- distinguish the relevant dimensions of available preservation
ploitation channels. objectives. According to di↵erent levels of complexity, role
and required domain knowledge the system o↵ers a subset
• Archiving and costs: The decision of following a spe- of questions which are well understood and the best avail-
cific institutional preservation policy for a given tech- able choice for a user to express his needs. Fig. 3 presents
nology is heavily influenced by given hardware and sample workflow which could be used to determine a given
budget constraints. Future exploitations on the costs user profile. For example a private user (ut = private per-
for content exploitation need to be predicted and taken son) with a solid level of IT knowledge (itk = expert) will
into account. be asked about preferred encodings and compression types
of the digital content, while others would define attributes
Other scenarios may include:
about storage limitations and upload samples of a given col-
• Provenance metadata lection.
• Data exchange and collaborative data enrichment 3.4 Recommendation computation
Di↵erently to classic KBRs where the application’s scope
• Publishing and digital rights management is very well delimited in terms of selecting the best match-
The definition of preservation dimensions is not orthogo- ing items in a list of known possibilities, the DiPRec sys-
nal and therefore certain properties might be involved more tem relies on expressing an institutional preservation con-
than once when computing di↵erent risk score. Due to text in form of user requirements that are combined with the
management and maintenance reasons properties are also knowledge acquired about the long term accessibility threat-
grouped by sets and a property may belong to one or more ening. We employ tools to evaluate the content of a given
property sets. The extent to which a property belongs to a collection from a technical point of view and to generate fine
grained preservation risk scores. When records are identified
15
http://sourceforge.net/projects/droid/ to have vulnerabilities on certain preservation dimensions a
55
the migration of the files available in T IF F/5 format to
JP EG/2000 by using the IM AGE M AGICK software with
standard settings.
4. EVALUATION
The evaluation of the first prototype of DiPRec was con-
ducted within the scope of the Assets project. Ten partners
of the project consortium provided metadata and binary
content (10 collections with a total size of 516GB contained
in 368067 media files) for supporting the development and
testing of services developed within the scope of the project.
The first step in the evaluation process was the identification
of file formats, definition of property sets and the aggrega-
tion of the domain knowledge available in open knowledge
bases on these file formats.
The Table 3 lists the distribution of file formats by content
type. Even the experimental data was taken from a small
number of content providers, we discovered a variety of 18
formats in 38 di↵erent versions used for encoding the digital
content.
Content Type File Format # Versions # Files
TEXT TXT 1 4
TEXT DOC 1 16
TEXT XML 1 20101
Figure 3: User requirements elicitation workflow TEXT HTML 1 1205
IMAGE JPG 8 323332
IMAGE PSD 1 3
rule based engine as JBoss Drools16 is used to propose ap- IMAGE PNG 4 1228
propriate preservation actions. The set of available business IMAGE BMP 2 141
rules are defined by domain experts in form of simple IF- IMAGE GIF 2 1066
IMAGE TIFF 4 4
THEN-ELSE rules. These rules are neither complete nor IMAGE PDF 16 25008
meant to be non-overlapping. Unified tool access for pro- AUDIO MP3 1 3634
cessing executable preservation plans is provided through VIDEO FLV 1 9468
the Assets preservation normalisation framework which is VIDEO MPEG4 1 935
able to invoke the tools with exactly defined settings and VIDEO MPEG1 1 3074
parameter configurations. VIDEO MPEG3 1 3074
3D PLY 1 50
3D DAE 1 307
IF ( rac > 0.5 AND ia == true AND iwa == true AND
open f ormat == FALSE)
THEN migrate(preservation f ormat) Table 3: Distribution of file formats in Assets col-
lections.
IF (content type == IMAGE)
THEN preservation format = (JPEG/2000:1, TIFF/6:0.8)
IF (f ile f ormat == TIFF/5 AND
The Digital Record Object Identification tool (DROID)
preservation f ormat == JPEG/2000)
version 5, signature file 45 was executed through the As-
T HEN migration tool = IM AGE M AGICK (2) sets preservation normalisation tool suite and was able to
successfully identify file formats in 95 percent of the cases
through its binary signature method except of the 3D model
The preservation recommendations are computed using
objects which have not yet been collected by Pronom. Ap-
the constraint solving problems (CSP) theory [8, 11]. Con-
propriate information on all of the file formats was contained
straints are defined within the preservation actions knowl-
in DbPedia and Freebase and the domain knowledge acqui-
edge base, the CSP context is defined by user profiles and
sition process was completed by successfully computing the
the preservation risks are identified for the given data col-
preservation risk analysis scores.
lection. The recommendations are represented in form of
The second part of the evaluation consisted in comput-
preservation actions. For example, the set of business rules
ing recommendations for the given content. Therefore, we
defined above combined with a user profile indicating inter-
created a user profile for content providers that are inter-
est in the dimension of archiving and web accessibility will
ested in making their content accesible through Europeana.
lead to the following recommendation when analyzing a col-
Within this context, the content providers manifest interest
lection of images in TIFF format:
for the web accessibility digital preservation dimension.
migrate(T IF F/5, JP EG/2000, IM AGE M AGICK)
The highest diversity of file formats was found in the
In free text translation, the recommendation will suggest
image collections. The recommendation to migrate these
16
http://www.jboss.org/drools files to the JPEG 2000 format didn not get a high priority
56
and will be performed within the next period of scheduled 6. ACKNOWLEDGMENTS
storage migration. The Image Magick tool was the recom- This work was partially supported by the EU project ”AS-
mended choise for performing this transformation action. SETS - Advanced Search Services and Enhanced Technolog-
The whole audio content available in Assets was provided in ical Solutions for the European Digital Library” (CIP-ICT
the mp3 format and no recommendation was made for trans- PSP-2009-3, Grant Agreement n. 250527).
forming audio collections. The most restrictive constraints
for web accessibility are defined for the video content. The 7. REFERENCES
pseudostreaming protocol is an advanced technological solu- [1] Aitken, B., Helwig, P., Jackson, A., Lindley, A.,
tion used for distributing information over the web. It allows Nicchiarelli, E., Ross, S.: The planets testbed: Science
the user to interact with the media-player and to quickly for digital preservation. Code4Lib 1(3) (2008),
navigate within the content without the needed to down- http://journal.code4lib.org/articles/83
load the entire media file. This protocol is supported by [2] Becker, C., Kulovits, H., Guttenbrunner, M., Strodl,
two file formats: flash video (FLV) and MPEG4 with H2.64
S., Rauber, A., Hofman, H.: Systematic planning for
video encoding. It has native support in HTML5 and is used
digital preservation: evaluating potential strategies
in HTML4 with an adequate browser plugin. A part of the
and building preservation plans. International Journal
Assets content is already available in FLV format and an- on Digital Libraries 10(4), 133–157 (2009),
other part is available in MPEG1 or MPEG2. The DiPRec http://dblp.uni-trier.de/db/journals/jodl/
resulting recommendation is to migrate the content to FLV
jodl10.html#BeckerKGSRH09
by using the ↵mpeg 17 tool.
[3] Becker, C., Kulovits, H., Rauber, A., Hofman, H.:
Plato: a service-oriented decision support system for
5. CONCLUSION preservation planning. In: Proceedings of the 8th
Within this paper we introduced the DiPRec recommender ACM/IEEE-CS joint conference on Digital libraries.
system, an expert support tool in the domain of digital long- pp. 367–370. ACM, New York (2008), http:
term preservation for GLAMs. An important contribution //publik.tuwien.ac.at/files/PubDat_170832.pdf,
of this papers is the exploitation of an open linked data ap- vortrag: 8th ACM/IEEE-CS joint conference on
proach for constructing the recommender’s knowledge base Digital libraries (JCDL 2008), Pittsburgh,
built upon open registries as DbPedia and Pronom. Since Pennsylvania; 2008-06-16 – 2008-06-20
the knowledge acquisition, aggregation and unification pro- [4] Becker, C., Rauber, A.: Four cases, three solutions:
cess is fully automated it is easy to upgrade the recom- Preservation plans for images. Tech. rep., Vienna
mender’s knowledge base. University of Technology, Vienna, Austria (April 2011)
We looked at preservation planning which is the process of [5] Becker, C., Rauber, A., Heydegger, V., Schnasse, J.,
specifying clearly defined and relevant trees of objectives in a Thaller, M.: A generic xml language for characterising
defined preservation dimension and evaluating them within objects to support digital preservation. In: SAC ’08:
a given (institutional) context to generate well-documented Proceedings of the 2008 ACM symposium on Applied
decisions. DiPRec is able to advance the process with in- computing. pp. 402–406. ACM, New York, NY, USA
ferred community knowledge and reduces the degree of man- (2008)
ual evaluation processes or require technical expertise in this [6] Bizer, C., Heath, T., Berners-Lee, T.: Linked data -
process. the story so far. Int. J. Semantic Web Inf. Syst. 5(3),
Am important concern related to the KBRs is the trust 1–22 (2009)
in the provided recommendations. This is especially rele- [7] Burke, R.D.: Hybrid web recommender systems. In:
vant for the digital preservation domain where we deal with The Adaptive Web. pp. 377–408 (2007)
a large amount of multimedia material and the execution [8] Felfernig, A., Gordea, S.: Ai technologies supporting
of the preservation actions is associated with considerable e↵ective development processes for knowledge-based
costs. Within this paper we did not examine the complete- recommender applications. In: SEKE. pp. 372–379
ness, correctness and quality degree of the underlying data. (2005)
We however argue that data from open knowledge bases like [9] Heitmann, B., Hayes, C.: C.: Using linked data to
DbPedia or Freebase could protect from biases introduced build open, collaborative recommender systems. In:
by the economical interests of professional companies by its In: AAAI Spring Symposium: Linked Data Meets
underlying community approach. Artificial IntelligenceŠ. (2010
The tool has been designed by reusing our past experience
[10] Heitmann, B., Hayes, C.: Enabling case-based
in building knowledge based and case based recommender
reasoning on the web of data. In: The WebCBR
systems [23, 8] and combining it with the expertise of cre-
Workshop on Reasoning from Experiences on the Web
ation long-term preservation infrastructure and applications
(2010)
[14, 1]. Based on this work the Assets normalisation tool
[11] Jannach, D., Zanker, M., Fuchs, M.: Constraint-based
suite is able to automate the process of object identification
recommendation in tourism: A multiperspective case
and characterisation and therefore directly integrates within
study. J. of IT & Tourism 11(2), 139–155 (2009)
the property evaluation, risk analysis and recommendation
process for a given record. We presented a first evaluation [12] Jannach, D., Zanker, M., Jessenitschnig, M., Seidler,
of digital content provided by national libraries and archives O.: Developing a conversational travel advisor with
through the Assets project where the underlying concepts of advisor suite. In: ENTER’07. pp. 43–52 (2007)
the DiPRec approach were proven to work adequately. [13] Jens, L., Jörg, S., Sören, A.: Discovering unknown
connections -the dbpedia relationship finder. In:
17
http://www.ffmpeg.org/ Proceedings of the 1st Conference on Social Semantic
57
Web (CSSW). vol. P-113, pp. 99–109. Gesellschaft für
Informatik, Leipzig, Germany (2007)
[14] King, R., Schmidt, R., Jackson, A., Wilson, C., Steeg,
F.: The planets interoperability framework: An
infrastructure for digital preservation actions. In:
ECDL09 Proceedings of the 13th European conference
on Research and advanced technology for digital
libraries. vol. 5714/2009, pp. 425–428. Springer-Verlag
(2009),
http://dx.doi.org/10.1007/978-3-642-04346-8_50
[15] Kurt, B., Colin, E., Praveen, P., Tim, S., Jamie, T.:
Freebase: a collaboratively created graph database for
structuring human knowledge. In: SIGMOD ’08
Proceedings of the 2008 ACM SIGMOD international
conference on Management of data. pp. 1247–1249.
ACM, New York, NY, USA (2008)
[16] Lindley, A., Jackson, A.N., Aitken, B.: A collaborative
research environment for digital preservation - the
planets testbed. Enabling Technologies, IEEE
International Workshops on 0, 197–202 (2010)
[17] Nonaka, I., Takeuchi, H.: The Knowledge-Creating
Company: How Japanese Companies Create the
Dynamics of Innovation. Oxford University Press
(May 1995)
[18] Orit Edelstein, Michael Factor, R.K.T.R.E.S.P.T.:
Evolving domains, problems and solutions for long
term digital preservation. iPRES 2011 - 8th
International Converence on Preservation of Digital
Objects (2011)
[19] Ricci, F., Werthner, H.: Case base querying for travel
planning recommendation. Journal of IT & Tourism
4(3-4), 215–226 (2001), http://dblp.uni-trier.de/
db/journals/jitt/jitt4.html#RicciW01
[20] Shabir, N., Clarke, C.: Using linked data as a basis for
a learning resource recommendation system. In: 1st
International Workshop on Semantic Web
Applications for Learning and Teaching Support in
Higher Education (SemHE’09) (September 2009),
http://eprints.ecs.soton.ac.uk/18053/
[21] Strodl, S., Becker, C., Neumayer, R., Rauber, A.: How
to choose a digital preservation strategy: evaluating a
preservation planning procedure. In: JCDL ’07:
Proceedings of the 2007 conference on digital libraries.
pp. 29–38. ACM, New York, NY, USA (2007),
http://doi.acm.org/10.1145/1255175.1255181
[22] Sven Schlarb, Edith Michaelar, M.K.A.L.B.A.S.R.A.J.:
A case study on performing a complex file-format
migration experiment using the planets testbed. IS&T
Archiving Conference 7, 58–63 (2010)
[23] Zanker, M., Gordea, S., Jessenitschnig, M., Schnabl,
M.: A hybrid similarity concept for browsing
semi-structured product items. In: EC-Web. pp. 21–30
(2006)
[24] Zanker, M., Jessenitschnig, M., Jannach, D., Gordea,
S.: Comparing recommendation strategies in a
commercial context. IEEE Intelligent Systems 22(3),
69–73 (2007)
58