TICTOCRON AN AUTOMATIC SOLUTION FOR PROPAGATING QUALITY METADATA TO

tictocron: an automatic solution for propagating quality metadata to scholarly toc rss feed metadata santiago chumbe1 and roddy macleod2

TICTOCRON: AN AUTOMATIC SOLUTION FOR PROPAGATING QUALITY METADATA TO
SCHOLARLY TOC RSS FEED METADATA
Santiago Chumbe1 and Roddy MacLeod2
1,2Heriot Watt University, UK
[email protected], [email protected]
Abstract
Institutions and researchers stand to benefit from the facilitation of
more widespread syndication of, and easier access to, Table of Content
(TOC) RSS (Really Simple Syndication [1]) feeds produced for scholarly
journals. However, many journal TOC RSS feeds are at present being
produced with erroneous, poor or incomplete metadata. This can hamper
the usefulness of scholarly current awareness services, and also cause
problems for individual subscribers to those feeds. This is exactly
what the ticTOCron software toolkit aims to overcome. The ticTOCron
toolkit automatically enhances poor, heterogeneous and incomplete
metadata found in TOC RSS feeds by making use of a pre-defined "Best
Practice" metadata scheme suitable for scholarly journals. In this
work we depict the main issues and "bad practices" found in TOC RSS
metadata obtained from more than 435 scholarly publishers. Then, we
describe software solutions implemented via ticTOCron. Some references
are made to the algorithms for generating semantic relations within,
between and from the harvested TOCs and to the mechanisms for
propagating "metadata associations" from a previously crawled
metadata-rich reference set. However, an effort is made to avoid
technical jargon and to replace complex technical descriptions with
samples and simple comparisons. The original metadata is converted to
a canonical format using the "Best Practices metadata set" for
scholarly papers proposed by the ticTOCs Project [2]. We also present
the results produced by ticTOCron when it was used for enhancing and
normalizing TOC RSS feeds collected from more than 12,000 journals.
Finally we propose a sustainable and scalable computational model
whereby the automatic solution is complemented and fine-tuned by a
cost-effective human cross-validation process.
Keywords:
Metadata quality enhancement, Journal TOC RSS feeds, Current
Awareness, CRON job
1. Introduction
Keeping up-to-date with the latest publications is at the heart of
much research activity. Publishers, librarians, authors and readers
have an interest in ensuring better exploitation and early access to
the latest output research. That is why considerable effort is being
expended by a growing number of publishers on providing RSS feeds for
their journal TOCs.
Using RSS readers, users can view details of the latest articles as
soon as they have been published, without having to visit individual
publishers’ websites where the TOC has been taken from. Clearly, all
stand to benefit from the facilitation of more widespread syndication
and easier access to TOC RSS feeds. RSS readers and current awareness
services that merely propagate what publishes have put in their RSS
feeds do not solve the key problems that researchers face when using
TOC RSS feeds to discover up-to-date knowledge from an avalanche of
ever increasing research output.
Achieving an automatic mechanism to serve the needs for currency of
this growing number of users has been always challenging. Online
current awareness services are important to virtually all librarians
and researchers as has been noticed by Farooq [Farooq, U. et al 2008].
As part of this process, users have a growing interest in the RSS
feeds that publishers are providing for an increasing number of
scholarly journal’s TOCs. However, we have found that the vast
majority of publishers are including erroneous, insufficient or poor
quality metadata in their TOC RSS feeds. The reasons why publishers
produce poor quality TOC RSS feeds can vary from lack of knowledge
about the technology or its benefits, to being reluctant to implement
expensive m
*
More than 6 million researchers worldwide
*
Around 1.5 million articles are written annually
*
About 24,000 scholarly journals
*
2,000 publishers publishing these scholarly journals
*
The number of published journals and researchers increases by
about 3% per annum.
*
More than 16,000 journals provide TOC RSS feeds
(Various sources, including the International Association of
Scientific, Technical and Medical Publishers)
etadata quality assurance systems.
To help to solve or mitigate this problem, ticTOCs on one hand has
prepared and proposed recommendations for publishers to understand the
importance of providing quality metadata in their feeds. On the other
hand ticTOCs has developed a semi-automatic mechanism for augmenting
the quality of metadata harvested from TOC RSS feeds. The combined
effect of these two works has enabled the prototyping of a web-based
environment that allows researchers to discover, subscribe to,
personalise, export and reuse TOC RSS feeds. It also supports the
creation of APIs (Application Programming Interfaces) to facilitate
the re-use of aggregated journal TOC content on a subject or community
basis by gateways, subject-based resource discovery services, library
services, portals and other services.
Section two of this paper will discuss the deficiencies identified in
the TOC RSS feeds. Most of those problems, such as parsing erroneous
XML files can be solved by using well-known software tools such as the
Universal Feed Parser [Lerner, 2004] However, the lack or
insufficiency of desirable information in the TOC RSS feeds is a
problem that so far has not been solved systematically. RSS feeds are
usually just consumed but not augmented by RSS readers. In this paper
we describe a solution that combines metadata crawled from the web
with normalization of metadata extracted from TOC RSS feeds. We use
content found on the web to complete the “missing” information in the
TOC RSS feeds. In fact what we have done is to generate “new” metadata
from related pre-existing metadata sources. In particular our
development has been inspired by the work done by Lagoze [Lagoze et
al, 2006] who has reported using web crawler-based metadata to augment
the quality of poor OAI (Open Archives Initiative) metadata with
significant positive results.
In the next sections we describe the algorithm and the composition of
the ticTOCron toolkit used by the Project to automatically collect
(harvest) TOC RSS feeds and augment their quality. We conclude by
discussing the impact of ticTOCron on the quality of TOC RSS feeds,
its possible applications for supporting or enhancing the work of
specific communities of users, and finally we draft some conclusions.
2. Problems with Journal TOC RSS Feeds
There are problems associated with the technology used to produce the
feeds. Other major problems are associated with the quality and the
content of the feeds. Although more than 80% of TOC RSS feeds provide
the TOC for the latest issue of the journal, the variations in the
quality and type of content are vast, and range from simple alerts
that a new issue is available, to whole back-catalogues of issues. We
have classified those problems in three main categories.
2.1. Technology Barriers
RSS is another example of theoretical standards undermined by chaotic
reality of the web. To make things worse, RSS has many competing
standards (in practice, Atom is just another syndication standard
[Hammersley, 2005]). There are two main branches of RSS: RSS 1.0 and
RSS 2.0 [Hammersley, 2005]. RSS 2.0 format is simpler than RSS 1.0,
which can be extended by the use of modules [Wittenbrink, 2005] to
provide rich metadata. We found out that only two publishers were
exclusively using Atom to provide TOC feeds. Nevertheless, handling
different syndication formats is no longer an issue for service
providers.
Feed Standard
Journals
%
RDF Module
Journals
RSS 2
6,598
53.7
- (no modules)
4,506
RSS 1 (RDF)
5,626
45.6
dc (Dublin Core)
3,433
Atom
6
0.0
dc prism
3,966
XML files
91
0.7
dc prism content
401
Total:
12,321
100.0
Table 1. Number of journals using different syndication formats (as
Jan 2008.)
Table 2. Number of journals using RDF modules recommended by ticTOCs
(as Jan 2008.)
Invalid and non well-formed TOC RSS feeds are also a common issue.
Thus, it is quite common to find feeds that use HTML mark-up within
the title or description tags. Some publishers are not aware of
modules such as content:encoded, which is a suitable way to display
HTML including hyperlinks to images. In general, publishers are
failing to notice that the metadata included in their TOC RSS feeds
have the potential of being more than simple headlines. Without a
focus on re-use, TOC RSS metadata is less useful. With that aim, we
have encouraged publishers to use a set of available RDF modules [5]
to enrich their feeds. Currently there are 3 official modules and 19
proposed modules [Manola et al, 2004]. Any syndication format can be
enrich with RDF modules, however use of RSS 1.0 would be advisable as
it follows the RDF specifications. The PRISM, Syndication, Content and
Dublin Core modules [5] have been identified as suitable modules for
TOC RSS feeds [Rogers, 2008].
However, as RSS 1.0 is extensible by design, modules can be written by
anyone, which can unnecessarily produce heterogeneous metadata
encodings. For example, the following “unsuitable” modules were found
in some TOC RSS feeds collected from publishers: CC, FEEDBURNER, FOAF,
and HTML. Yet, it was encouraging to notice that by Feb 2009, more
than 2,000 TOC RSS feeds have started to include elements from the
Syndication module in their feed’s headers. This is a positive
development. The Syndication module has the potential to enable
toolkits such as ticTOCron with the capacity of performing selective
harvesting, which would save computational time and make it possible
to update only the new TOC RSS feeds and as soon as they have been
published.
2.2. Metadata tagging inconsistencies
There is a widespread difference in what publishers are putting in
their feeds. For further information there are the deliverables of the
ticTOCs Project [2].
The conclusion expressing that “you cannot trust the metadata provided
by any OAI data provider” [Chumbe, 2006] also applies to TOC RSS feeds
providers. The values that publishers are including in metadata
elements could be almost anything, regardless of the original purpose
of the element specified by metadata standards. For example Elsevier
put the authors of the article in the element;
Springer-Verlag uses to store the title, abstract,
authors, vol, issue, year, DOI, etc.; Biomed Central puts all the
authors in a single element; and IoP uses a
element for each author.
- multiple authors – IOP:
Luciano Telesca
Antonio Lanorte
Rosa Lasaponara
- multiple authors - Biomed Central:
Alexandra Devine, Michelle Kermode, and Helen
Herman
- multiple authors – Elsevier:
Dostal, M. , Roberts, J.B. , Holmes, R.
- multiple authors – Springer-Verlag:
article title, abstract, authors, Vol, Issu,
Year,
ISSN, etc..
In general, any value provided in the metadata is a potential source
of error. Odd values have been identified in the feeds, such as
“2086”. CrossRef [3], a main partner of ticTOCs is
expected to play a leading role in encouraging every publisher “to
play ball according to the rules.”
2.3. Feeds Content inconsistencies
Describing all the inconsistencies found in the content of TOC RSS
feeds would take too long. The inconsistencies are varied. Some
journals, such as the “Drugs in R & D” Journal, puts all the metadata
in the RSS feed header. Other journals use different types of formats
to register dates. The following is a summary of common
inconsistencies detected in the content included in TOC RSS feeds.
*
Various TOCs in the same feed. For ticTOCs, being a current
awareness service, the back-catalogue of issues previously
published in a journal is irrelevant, and this information might
also confuse individual RSS feed subscribers.
*
Some TOC RSS feeds contain other, different types of content. Not
only TOCs, but also News, Job Announcements, Editorials, etc.
*
Very few journals use the proper metadata elements to register
information on Vol, Issue or Year. A significant number of feeds
do not include this information at all.
*
Because in RSS feeds there isn’t such a concept as “mandatory” it
is quite common to find feeds without enough basic information
such as the title’s item and the link to its webpage.
UTF-8 encoding errors are also common. RSS Feeds generated by software
can be an obstacle for automatic harvesting. Thus
https://pi.library.yorku.ca/ojs/index.php/soi/feed/rss redirects
harvesters to an authentication webpage. The above examples are just
some cases illustrating poor metadata being exposed by TOC RSS feeds.
3. The ticTOCron toolkit
The aim of the toolkit is to automatically harvest TOC RSS feeds and
to use a suite of software tools to handle the problems affecting TOC
RSS feeds as identified in the previous section. The toolkit works as
an "invisible infrastructure" for the ticTOCs public service
prototype. ticTOCron itself is a set of software tools managed by a
computer system utility (CRON job) found in most UNIX systems [Aeleen,
2002].
The composition of the ticTOCron software is based on an adaptation of
the classification proposed by Berson [Berson, A. et al 2007] to
handle the gathering of metadata that we need to consume but for whose
quality we don’t have any influence. Berson classifies in five
categories the tools that can partially or completely automate the
tasks associated with extracting, cleaning, standardizing and
enriching metadata obtained from external sources:
- Auditing Tools
- Data Cleansing Tools
- Data parsing and standardization Tools
- Data extraction, transformation and loading tools
- Hybrid Packages
Each of the ticTOCron tools is a modular software component that has
been developed using a loosing coupling design technique. Each of the
tools has a specific goal as outlined by Bearson. Therefore, the
elimination or failure of any of these tools does not bring the entire
CRON system down. The sequence in which they are executed is
computationally irrelevant, but it has an impact on the consistency of
the results produced by the CRON process.
The tools produce logs and exception errors that are used by the
system administrator to solve errors and more importantly to increase
the “know-how” of the tools.

Figure 1: ticTOCron toolkit system context
3.1 Harvesting tool
There is a large number of software tools for parsing RSS feeds but
most of them will run into problems associated with invalid and/or non
well-formed XML files. Those RSS readers handle faulty feeds with
different levels of sophistication. Some of them will attempt to
correct the faulty feeds, but others will simply return an error in
the user interface. However a basic requirement for our harvester was
that it should be able to parse even non well-formed XML files and it
should not be stopped by unpredictable XML content. Compliance with
RSS standards is not taken into account at this stage. The tool will
accept any XML file and if the file is not a valid RSS feed, will
intend to identify any TOC-like structure in the file and try to
extract metadata. If the process fails, the harvester will log the
event and continue parsing the next RSS feed.
Select journals that haven’t been updated in the last week, starting
with the ones that are more frequently updated or their frequency is
unknown
For each selected journal
{
Check frequency of publication if available
If it needs to be updated or frequency is not available
{
Harvest journal TOC RSS feed
If there is a problem, log error and continue with next journal
Generate and update metadata for the journal feed (e.g. RSS version,
used modules, frequency of publication, etc.)
Parse feed to compare its contents with what is stored in the database
If contents have not changed continue with next journal
Remove items from previous feed
For each new item
{
Save item data
Generate and store additional common metadata, such as
quality level (0=need normalization, 1=need cleansing,
2=need auditing,
3=need cleansing & auditing. Default is 0)
}
}
}
Figure 2. ticTOCron Harvester Tool Algorithm
As most of the journal feeds do not provide the frequency of
publication, the harvesting process takes a long time to complete. The
method is still far from perfect because in order to succeed it
requires having a long enough harvesting history for each journal. The
computing time is inversely proportional to the number of feeds whose
frequency of publication is known.
Despite its importance, the automatic collection or harvesting of RSS
feeds is the easiest task done by the CRON job. The challenging part
starts after the metadata has been collected and deposited in the
database.
3.2. Data Cleansing Tool
This utilises a set of rules that have been setup in the software to
remove, combine or move harvested values to improve the quality of the
metadata and potentially to add new accurate content. It deals with
the TOC RSS items whose quality level is one (please see Fig. 2.)
Basically, the Cleansing tool analysis the value of each metadata
element with two aims; firstly to remove irrelevant items to leave
only the ones of the current issue; and secondly, to move or/and
create metadata from the harvested information. Thus, the tool will
remove from the TOC all the items that have unsuitable information
such as news, job announcements as well as old issues. Then it will
try to identify relevant missing information from each metadata
element for each TOC item.
For example if the Cleansing tool detects the presence of a valid DOI
(Document Object Identifier) in a metadata element, it will create a
new metadata element for the DOI. In most of the cases, the tool can
identify values commonly used for citing a scholarly journal article.
However, the tool will be unable to detect relevant information put by
the publisher in the wrong metadata element if none of its rules is
matched by contents of this element. For example we have noticed that
the feeds using the PRISM module are the ones that require little
cleansing work. The most challenging cases are produced by feeds that
store different types of useful information in the wrong element. An
extreme case is illustrated by some TOC RSS feeds produced by
Springer-Verlag. The description field of those feeds contains the
article abstract plus all the information that is required to create
the full article citation (Journal Title, ISSNs, article title,
abstract, authors, Vol, Issue, publication year, URLs, DOI, etc.) and
all of them enclosed by HTML tags.
The Cleansing tool analyses all the available information available in
the TOC items looking for additional information to augment the
metadata quality and richness. It is also in charge of handling the
correct metadata encoding (UTF-8 conversion) as well as of removing
html mark-up and spurious tags from the TOC items.
3.3. Auditing Tool
This tool compares the metadata of each TOC RSS item against the
item’s metadata found in the publisher website. The objective is to
enhance the accuracy of the RSS metadata and its correctness by
contrasting the metadata values of each item of the TOC RSS feed
against information crawled from the item’s web page. The success of
the Auditing tool depends on its ability to identify, in the item's
web page, metadata that is better than the TOC RSS metadata. This
factor in turn depends on the tool’s algorithm and the structure of
the web page that has been crawled and is being analysed.
There is a wide range of software tools for creating metadata
automatically from crawled web pages [Albassuny, 2008; Liu, 2007]. Our
tool is based on heuristic models that use the formatting and
composite structure of web pages rendering scholarly articles metadata
to break the ambiguousness of HTML documents [Tonkin, 2008].
Additionally, taking into account that extraction algorithms could
contribute to useful automatic metadata generation [Greenberg, 2003],
the Auditing tool uses a standard HTTP GET request to download the
item’s webpage and utilises its own TOC RSS metadata to create
associations between the crawled metadata values and the metadata
values obtained from the RSS feeds; then the problem of extracting
metadata from the web can be reformulated as one of
inter-extrapolation of values between a previously crawled
metadata-rich reference set and a related metadata-poor source (RSS.)
The knowledge gained in each extraction is stored and used as a
“reference data model” for the next time that the “same” web page is
parsed.
The amount of rich metadata propagated at the end of the auditing
process depends significantly on the quality and quantity of
associations that the Auditing tool has been able to create between
the crawled metadata and the RSS feeds’ metadata. We have achieved
good results for only a small number of selected journals. The tool
was not applied to all the journals because its software was still
under development when the ticTOCs Project was completed. Being
“metadata extraction” an inherently uncertain method and prone to
propagate errors, the tool only augments metadata if a 100% matching
between web-crawled and RSS feed metadata key values (e.g. article’s
title) is obtained. The initial results have shown that the auditing
tool effectiveness can be increased by allowing human intervention to
resolve the cases when no 100% matching was achieved. This is in
agreement with Liu’s findings [Liu, 2007] that not all metadata can be
generated automatically. Applying fuzzy logic models to express or map
the human’s semantic for the “matching” concepts into an exact numeric
range of acceptance or rejection is an approach being explored to
enhance the chances of identifying logic units from HTML document
sections.
3.4. Data Mapping and Normalization Tool
This tool combines the principles of Berson’s third and fourth tools
(Data parsing and standardization and, Data extraction, transformation
and loading tools.) The tool first atomizes the TOC RSS item metadata
elements in data units that are subsequently compared (mapped) against
the “ticTOCs Best Practices” metadata set proposed and recommended to
publishers by the ticTOCs Project [Rogers, 2008]. This “Best
Practices” metadata format includes the standard RSS 1.0 modules dc,
prism, content and syndication.
This tool also “normalizes” the cleansed and audited metadata in the
sense that it produces uniform, consistent, valid and well-formed
metadata values for each journal feed item and for all the harvested
journal feeds. Thus the original metadata is converted to a canonical
format which subsequently will be used by the service to expose TOC
RSS feeds via APIs.

Figure 3: ticTOCron tools outline
4. Impact of ticTOCron
The first positive impact on a TOC RSS feeds service as a result of
using ticTOCron is that its content is continuously being updated and
consistently exposed to third parties. Additional impacts may become
apparent as the usage of ticTOCs grows. The long term impact of our
work will derive from the response it receives from TOC RSS feeds
services, and its reuse by, or incorporation in, other related
software systems. In the next section we explore possible applications
that would be built using APIs to reuse the output generated by
ticTOCron.
A clear benefit of using ticTOCron is that once a feed has been
processed by ticTOCron, it will be available even when the publisher
RSS site is temporarily unavailable or it is producing XML errors. In
those cases the live feeds would not work, but the ticTOCs "cached"
version will still be providing users with access to the online
content of the TOC.
The second benefit is the richness of the metadata produced by
ticTOCron. Most of the feeds provide titles of the articles as well as
abstracts. However the abstracts are not always "abstracts."
Additionally, although the title and abstract would be enough to let
users quickly decide whether or not the article is relevant for them,
they wouldn’t be enough, for example, for creating citation lists.
Thus, more metadata is required to create full identifications or
citations (e.g. Authors, Vol, Issue, Date, etc.) By providing richer
metadata, ticTOCron enables easy creation of bibliographic references
and can indirectly make users more willing to visit the articles’
website.
Providing uniform, consistent, valid and well-formed metadata values
for all the journal feeds can also impact on the way in which other
services or users could create persistent/stable URLs to articles from
the TOC RSS feeds. As publishers tend to use different formats and
structures for publishing the content of their articles, laborious
work needs to be done by the user to create those URLs. For example,
when a DOI has been included in the article abstract, the user would
need to extract it manually to use in other applications such as
reading lists. Enabling services such as openURL resolution from TOC
RSS feed items is also made challenging by incomplete and inconsistent
metadata. In short we believe that the work done with ticTOCron would
increase the interoperability of TOC RSS feeds based services.
5. Possible Applications
Most of the possible applications of ticTOCron are connected with
reusing what ticTOCron has produced to create multiple outputs,
applications or services from that single source of journal TOCs
content. The standard, consistent and rich metadata produced by
ticTOCron allows service providers to single source journal TOC
content in powerful ways. The potential of ticTOCron output would be
increased by making small adjustments to its software or setup, such
as keeping previous harvested feeds, as a consequence, past issues
wouldn’t been overwritten when a new issue is detected by the
harvester (ticTOCs being a current awareness service, doesn't retain
all items in any aggregated database. Thus, each time a new journal
issue is published, items of the previous issue are removed.) To
increase the reusability of ticTOCs, the service would need to retain
items from “n” previous issues. The value of “n” is in function of the
frequency in which the issues of a journal are published and could be
used to calculate the number of issues that need to be retained for
each journal.
Another feasible application would be to create a service or facility
to generate a localised version of ticTOCs for an Institutional
Library. The advantage of ticTOCs for this purpose would be that
ticTOCs is likely to be more up to date than any other database, such
as ZETOC (Electronic Table of Contents, http://zetoc.mimas.ac.uk),
SCOPUS (http://www.scopus.com) or similar.
Once there is enough back-file data, a ticTOCron-based database would
be used to produce information on “research tendencies”, to answer
queries such as "in which subject or discipline have authors been
writing in the last six months" or "what are the recent trends in
recent research in my area of interest."
An application in the field of Institutional Repositories (IRs) would
be a facility to alert IR managers when a submitted paper from their
IR has been published in a scholarly journal. IR managers need to know
when the peer reviewed version of a paper is available. The fact that
the TOC RSS metadata has been augmented by ticTOCron makes TOC RSS
feeds a viable solution to help IR managers because it is expected
that the number of matches between “submitted” versions (from the IR)
and “published” versions (from ticTOCs) has a direct relationship with
the quality of the metadata produced by ticTOCron.
6. Conclusion
This research indicates that generators using both extraction and
harvesting methods have the potential to create useful metadata from
TOC RSS feeds. Although the results obtained are preliminary as the
ticTOCs service is new, the results show real possibilities of success
and draws attention to important areas of metadata generation
practice.
ticTOCron has been able to enhance the TOC RSS feeds for most of the
12,321 journals in ticTOCs. Although more work is needed in order to
make further use of the Auditing Tool, ticTOCron has shown itself to
be a potential tool to facilitate interoperability of TOC RSS feeds.
The process of enhancing TOC RSS feeds cannot be completely automated.
As in any application of technology for generating metadata, the best
results are achieved when humans are able to “teach” the software as
part of a continuous iterative process. This conclusion is in
agreement with what researchers have concluded that the most effective
mean of metadata creation is to integrate both human and automatic
methods [e.g., Schwartz, 2002; Craven, 2001]. As it is shown in Figure
3, human intervention occurs in all the stages of ticTOCs application.
The result is still a sustainable and scalable computational model
because ticTOCron uses “human intervention” to learn from “experience”
(TOC RSS feeds have structural patterns that are unlikely to change
frequently.) Consequently the needs for human intervention and
computational resources should decrease with the time.
A specific conclusion from this study is that using cleansing,
auditing, mapping and normalizing tools, derived from the Berson’s
model, is a suitable approach for creating optimal TOC metadata and
for augmenting the quality of harvested TOC RSS feeds.
The study also has confirmed that a correct use of RDF modules [5] can
be a key ingredient for any successful service built on top of a
database of TOC RSS feeds. The benefit of using those modules would be
enhanced by following simple best practices for interoperability, such
as do not restrict access to TOC RSS feeds, produce regular static
copies of RSS feeds for fast processing and make available up-to-date
OPML [6] files, etc.
Acknowledgements
----------------
ticTOCron R & D is based upon work supported by ticTOCs, a JISC funded
project under a “Users & Innovation: Personalising Technologies”
grant. The authors acknowledge the generous support of the entire
ticTOCs team. The methodologies, findings and conclusions described in
this manuscript are those of the authors and do not necessarily
reflect the views of JISC or ticTOCs.
References
----------
Aeleen Frisch, A. (2002) Essential System Administration: Help for
UNIX System Administrators. Published by O'Reilly, ISBN 0596003439,
97805960034321149, Third Edition. pp. 90-99
Albassuny, B.M. (2008) Automatic metadata generation applications: a
survey study. Int. J. Metadata, Semantics and Ontologies, Vol. 3, No.
4, pp.260–282.
Berson Alex, Dubov Larry and Dubov Lawrence. (2007) Master Data
Management and Customer Data Integration for a Global Enterprise. Book
published by McGraw-Hill Professional, ISBN 0072263490, 9780072263497.
406 pages
Craven, T., (2001) Changes in Metatag descriptions over time. First
Monday, Vol. 6, No. 10 - 1 Oct. 2001
Chumbe, S., MacLeod, R., Barker, P., Moffat, M. and Rist, R. (2006)
Overcoming the obstacles of harvesting and searching digital
repositories from federated searching toolkits, and embedding them in
VLEs. Proceedings 2nd International Conference on Computer Science and
Information Systems, Athens, Greece.:
http://eprints.rclis.org/archive/00006394
Farooq, U., Ganoe, Craig H., Carroll, John M., Councill, Isaac G., and
Giles, C. Lee (2008). Design and evaluation of awareness mechanisms in
CiteSeer. Information Processing and Management, 44, pp. 596–612
Greenberg, J. (2004). Metadata Extraction and Harvesting: A Comparison
of Two Automatic Metadata Generation Applications. Journal of Internet
Cataloguing, 6(4): pp. 59-82
Greenberg, J., Spurgin, K. and Crystal, A. (2006). Functionalities for
automatic metadata generation applications: a survey of metadata
experts’ opinions. Int. J. Metadata, Semantics and Ontologies, Vol. 1,
No. 1, pp.3–20.
Hammersley, B (2005) Developing feeds with RSS and Atom. Book
published by O'Reilly, ISBN 0596008813, 9780596008819. 276 pages
Lagoze, C., Krafft, D., Cornwell, T., Dushay, N., Eckstrom, D.,
Saylor, J. (2006) Metadata aggregation and "automated digital
libraries": a retrospective on the NSDL experience. Proceedings of the
6th ACM/IEEE-CS joint Int. Conference on Digital Libraries. pp.
230-239
Liu, J. (2007) Metadata and its Applications in the Digital Library:
Approaches and Practices. Publised by Libraries Unlimited, London,
pp.143–149.
Manola, F. and Miller, E. (2004) RDF Primer. W3C recommendation.
http://www.uazuay.edu.ec/bibliotecas/conectividad/pdf/RDF%20Primer.pdf
Margaritopoulos, M., Margaritopoulos, T., Kotini, I. and Manitsaris,
A. (2008). Automatic metadata generation by utilising pre-existing
metadata of related resources. Int. J. Metadata, Semantics and
Ontologies, Vol. 3, No. 4, pp.292–304.
Reuven M. Lerner, R., (2004) At the forge: aggregating syndication
feeds. Linux Journal, published by Specialized Systems Consultants,
Inc., ISSN: 1075-3583. Issue 128 (December 2004), Page 7.
Rogers, L. (2008) RSS and scholarly journal tables of contents: the
ticTOCs project, and good practice guidelines for publishers. FUMSI
Magazine, October 2008 [online] URL:
http://web.fumsi.com/go/article/share/3356
Schwartz, C. (2002). Sorting out the web: approaches to subject
access. Westport, Connecticut: Ablex publishing. Part of the
Contemporary Studies in Information Management, Policies, and Services
series by Hernon, P (Ed.)
Tonkin, E., Muller, H. (2008) Keyword and metadata extraction from
pre-prints. ELPUB2008. Open Scholarship: Authority, Community, and
Sustainability in the Age of Web 2.0 - Proceedings of the 12th
International Conference on Electronic Publishing held in Toronto,
Canada 25-27 June 2008. pp. 30-44
Van de Sompel, H., and Lagoze, C. (2001) The Open Archives Initiative
Protocol for Metadata Harvesting. URL:
http://www.openarchives.org/OAI_protocol/openarchivesprotocol.html
Wittenbrink Heinz. (2005) RSS and Atom: Understanding And Implementing
Content Feeds & Syndication. Packt Publishing. ISBN 1904811574,
9781904811572. 250 pages
Notes
[1] “RSS feeds” is a mechanism to efficiently deliver frequently
updated content to users (There is some discussion as to what RSS
stands for, but most people plump for 'Really Simple Syndication'. In
essence, the feeds themselves are just XML files, designed to be read
by computers rather than people).
[2] ticTOCs is a project funded by JISC to prototype a journal Tables
Of Contents (TOCs) current awareness service, available at
http://www.tictocs.ac.uk . ticTOCron has been developed as part of
ticTOCs.
[3] CrossRef is a not-for-profit association to enable easy
identification and use of electronic content (http://www.crossref.org)
[4] Document Object Identifier (http://www.doi.org)
[5] RDF RSS 1.0 Specification and Modules.
http://web.resource.org/rss/1.0
PRISM Module: http://www.prismstandard.org/resources/mod_prism.html
[6] OPML (Outline Processor Markup Language):
http://validator.opml.org