license
stringclasses
1 value
text
stringlengths
66
12.2k
doi
stringlengths
10
59
type
stringclasses
14 values
detected_lang
stringclasses
6 values
publication_year
float64
2.01k
2.02k
is_dataset
bool
2 classes
is_software
bool
2 classes
is_acknowledgement
bool
2 classes
is_clinicaltrial
bool
2 classes
field_name
stringclasses
26 values
field_id
stringclasses
26 values
cc-by
in collaboration with the South Australian Department of Health and Ageing conducted this research under National Health and Medical Research Council (NHMRC) projects (# 631917 and # 570150 ) investigating the relationships between place and metabolic syndrome (PAMS). Catherine Paquet was funded by a National Hea...
10.3390/ijerph13050519
article
en
2,016
false
false
true
false
Social Sciences
https://openalex.org/fields/33
cc-by
UnlimitID is a method for enhancing the privacy of commodity OAuth and applications such as OpenID Connect, using anonymous attribute-based credentials based on algebraic Message Authentication Codes (aMACs). OAuth is one of the most widely used protocols on the Web, but it exposes each of the requests of a user for da...
10.1145/2994620.2994637
preprint
en
2,016
true
true
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
The RPs also link the user to their IdP, preventing pseudonymous or anonymous usage of their service. We present UnlimitID, which uses lightweight attributebased anonymous credentials based on algebraic MACs [1] to provide unlinkability between an IdP and the various RPs for a given user using unlinkable, but multi-u...
10.1145/2994620.2994637
preprint
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
It is not necessary to link capabilities to a single identity. Previous work has discovered a large number of security vulnerabilities in OAuth 2. 0, but a more fundamental problem that has not been addressed is that the identity provider in OAuth is assumed to be completely trusted and so may log all user transactions...
10.1145/2994620.2994637
preprint
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
One is to change the flow so that data is sent from the IdP to the user and then from the user to the RP, as done by Mozilla BrowserID and SPRESSO [3] , achieving unlinkability but at the cost of requiring the client to be online for all transfer of attributes (unusable in practice) and moving the root of trust to the...
10.1145/2994620.2994637
preprint
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Protocol Flow We assume that the user has already been issued a credential for that epoch, as advised in 3. 3. 1. The flow is presented in Figure 1 and explained in detail in the following steps: 1. A user visits a RP that requires authorization or authentication. 2. The RP redirects the user to the IdP. 3. The user ...
10.1145/2994620.2994637
preprint
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
5. The IdP redirects the user back to the RP, along with an authorization code for accessing the respective pseudoidentity database entry. In case the pseudonym attribute was revealed, the user appends a signature under the public key UidService = P K = Gid Sub Service on the authorization code, which is kept fresh by ...
10.1145/2994620.2994637
preprint
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
PROTOTYPE PERFORMANCE We implemented the IdP issuing of credentials and pseudonym registration in Python to evaluate its performance 1. The implementation consists of 1874 lines of Python 3. 4 for the IdP credential server, divided into 495 lines for the IdP server, 601 lines of a generic zero-knowledge proof library a...
10.1145/2994620.2994637
preprint
en
2,016
false
true
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
All tests were performed on an Intel Core i7-4700MQ running at 2. 4Ghz with 16. 0GB of RAM. The numbers reported are an average of 10 runs. Timings were gathered using the standard library time. monotonic() clock and all times are reported in milliseconds (ms). We measured the timing of the full user & server computati...
10.1145/2994620.2994637
preprint
en
2,016
false
true
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
CONCLUSIONS UnlimitID demonstrates that the OAuth protocol can be upgraded so as to not violate user privacy by creating unlinkable pseudo-identities based on multi-show attribute-based credentials. Unlike PseudoID [2] and Crypto-Book [8] , Un-limitID is usable within current OAuth flows with modifications needed on...
10.1145/2994620.2994637
preprint
en
2,016
false
true
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Acknowledgements. The authors are supported by NEXT-LEAP ( EU H2020 ref: 688722 ) and EPSRC Grant EP/M013-286/1.
10.1145/2994620.2994637
preprint
en
2,016
false
false
true
false
Social Sciences
https://openalex.org/fields/33
cc-by
Secondly, using a single projection to explore the future may give decision-makers the impression that the direction of change is unique. This is not the case: several other projections exist and full uncertainty analyses could be completed using appropriate data sets. Predicting and anticipating the consequences of cl...
10.1186/s40322-016-0034-y
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
), greenhouse gas (GHG) emission reference scenarios, climate downscaling methods, etc. Even though all sources of uncertainty are not yet understood and measurable, it is important to start evaluating the uncertainty linked to projections produced by climate indices like tourism climate indices. This is illustrated he...
10.1186/s40322-016-0034-y
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Mieczkowski's indice and computation formulae: France The objective here is to show that the evaluation of suitable tourism conditions is sensitive to slight changes in computation methods. As an example, we focus on the Mieczkowski's indice. The study is performed for France, using the high resolution SAFRAN database ...
10.1186/s40322-016-0034-y
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
However, for the winter and summer seasons, which are of high importance for tourism, Fig. 2 shows that the "daily" method appears to produce far less spatially contrasted results than the "Mieczkowski" method. An assessment of long-term uncertainty in Mieczkowski's indice projections: using the ENSEMBLES database Tou...
10.1186/s40322-016-0034-y
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Acknowledgments This paper was written by a team participating in the EU FP7 project CLIM-RUN "Climate Local Information in the Mediterranean region Responding Users '.
10.1186/s40322-016-0034-y
article
en
2,016
false
false
true
false
Social Sciences
https://openalex.org/fields/33
cc-by
Description of participants/cohorts This project builds on two existing cohorts,the NuAge cohort in Canada, and the RECORD cohort in France-, and further launches a new cohort of older adults in Luxemburg. Canada: A sub-sample of 175 participants living in urban and suburban neighbourhoods of Montreal and Sherbrooke in...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Complementary physiological and biological data was further obtained through a 2-h medical examination, providing objective measures of height, weight, waist circumference, blood pressure, and lipid and glycemic profiles. The second wave, started in 2011, is ongoing, with additional participants recruited to compensate...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
In 2015, a novel project was funded by Agence Nationale de la Recherche to further extend data collection among a subsample of 400 older adults within RECORD, allowing collection of social network measures as partly done in the CURHA project. With this new project, comparability with the Canadian and Luxembourg cohorts...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Luxembourg: In Luxembourg, a new cohort is being developed for this project by the LISER Group. A target recruitment of 500 participants aged 65 and up is based on a sampling scheme using Social Security administrative files. These anonymized micro-data allow the selection of all elderly while accounting for some socio...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
VERITAS-social: collecting regular activity places and social networks We will administer VERITAS, an interactive map-based questionnaire designed to collect information on daily mobility, including identification of regular destinations and frequency of visit, perceptions of neighbourhood limits, and complementary dat...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
These questions allow assessing the emotional closeness between participants and alters, and identification of relationships between network members. Figure 3 illustrates both social and spatial network information collected through VERITAS- Social. Within the HANC protocol, VERITAS will be used to collect social ne...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
For this study, we will only use the GPS and accelerometer functions. For the RECORD GPS Study, an Actigraph GT3X and a QStarz BTX1000 GPS tracker were used. A number of data processing methods and algorithms have been developed in collaboration between Kestens' & Chaix' teams, specifically through work on the RECORD G...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
These specific empirical objectives are firstly analysed within the Canadian and Luxembourg national fieldworks and secondly developed as national outputs for cross-comparison between countries. The qualitative section of the research will be closely linked to both GPS tracking and quantitative surveys. As part of a qu...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Outcomes of this mixed-method approach will be threefold: (i) establish the meaning of mobility in terms of spatial uses and daily routines, and highlight cultural differences in perceptions of mobility/behaviors between three national fieldworks, and (ii) uncover the complex relations between subjective experiences in...
10.1186/s12877-016-0273-7
article
en
2,016
false
true
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Measures of interest Mobility assessment Raw 1-s GPS data obtained from the SenseDoc will be processed to derive detailed schedules of activity locations visited, trips, and estimations of transportation modes [84].
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Complementary environmental predictors derived from GIS In each setting, participant's place of residence as well as reported activity locations and GPS data will be linked to existing GIS platforms that contain rich information on urban contexts. Whereas a preliminary assessment has allowed identifying a series of dat...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Statistical analyses We will develop quantitative models integrating the collected data on daily mobility, social networks, physical activity and perception of places into geographic information systems to develop new indicators and spacetime modeling of health and place relations in regards to healthy aging. Mobility ...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Discussion This protocol offers key data collection components to unravel the complex interactions between daily mobility, social networks, urban environments, and healthy aging. Relevant data collected includes information on older adults' regular destinations, social network structure and in-person encounters, daily ...
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Availability of data and material The datasets generated through this study are deposited at the principal investigators' laboratories. Please contact us if you are interested in conducting analyses.
10.1186/s12877-016-0273-7
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Acknowledgments The team would like to thank all past and future participants of this study. The study was funded in Canada by the Canadian Institutes of Health Research (CIHR) and the Fonds Quebecois de la Recherche en Santé (FRQS) , in Luxembourg by the RECORD GPS Study which received funding from the Fonds Nat...
10.1186/s12877-016-0273-7
article
en
2,016
false
false
true
false
Social Sciences
https://openalex.org/fields/33
cc-by
Acknowledgments The team would like to thank all past and future participants of this study. The study was funded in Canada by the Canadian Institutes of Health Research (CIHR) and the Fonds Quebecois de la Recherche en Santé (FRQS) , in Luxembourg by the RECORD GPS Study which received funding from the Fonds Nat...
10.1186/s12877-016-0273-7
article
en
2,016
false
false
true
false
Social Sciences
https://openalex.org/fields/33
cc-by
Thirdly, the contrast between the United States and the European Union in their treatment of intellectual property in databases, with the United States retaining the integrity and scope of copyright (Feist, 1991) and the European Union developing sui generis provisions for databases is considered. In this instance, t...
10.1002/pra2.2016.14505301016
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Methods The study area extends throughout the Iberian Peninsula and the Balearic Islands (Fig. 1 ). The method used to reconstruct past climate variables is based on the probability density functions (PDFs) of plant taxa identified in fossil pollen records, and it requires a georeferenced distribution of modern plant t...
10.5194/cp-12-1137-2016
article
en
2,016
true
false
false
false
Earth and Planetary Sciences
https://openalex.org/fields/19
cc-by
Source is either the European Pollen Database (EPD) or author contribution. Longitude and latitudes correspond to the centroid of the nearest cell to the site and altitude as extracted from WorldClim data set, all at 5 spatial resolution. Each site has information about the number of 14 C dates available, the temporal ...
10.5194/cp-12-1137-2016
article
en
2,016
true
false
false
false
Earth and Planetary Sciences
https://openalex.org/fields/19
cc-by
Data sources The current distribution data for 246 taxa were obtained by georeferencing the Atlas of Florae Europaeae (Jalas and Suominen, 1972 , 1973 , 1976 , 1979 , 1980 , 1983 , 1986 , 1989 , 1991 , 1994; Jalas et al. , 1996 Jalas et al. , , 1999;; Laurent et al. , 2004). We gathered additional occurrence data for ...
10.5194/cp-12-1137-2016
article
en
2,016
true
false
false
false
Earth and Planetary Sciences
https://openalex.org/fields/19
cc-by
The global observed climate data for January minimum temperature (T jan ), July maximum temperature (T jul ) and annual precipitation (P ann ) data were obtained from WorldClim database (Hijmans et al. , 2005, www. worldclim. org) with 5 resolution (∼ 10 km). The climate data were downscaled to the same spatial resol...
10.5194/cp-12-1137-2016
article
en
2,016
true
true
false
false
Earth and Planetary Sciences
https://openalex.org/fields/19
cc-by
In order to evaluate the robustness of the reconstruction method, we have compared modern reconstructed and observed climate data from WorldClim database. For reconstructing climate from pollen data, we have used all samples available within the last 500 years. Climate values were averaged for all sites with more than...
10.5194/cp-12-1137-2016
article
en
2,016
true
false
false
false
Earth and Planetary Sciences
https://openalex.org/fields/19
cc-by
Climate stability was computed for each variable as the mean absolute deviance from the current climate as available in WorldClim data set. All analysis were performed using the R Project for Statistical Computing (R Development Core Team, 2012) with packages fields (Furrer et al. , 2012) , rgdal (Keitt et al. , 2012...
10.5194/cp-12-1137-2016
article
en
2,016
true
true
false
false
Earth and Planetary Sciences
https://openalex.org/fields/19
cc-by
Data availability The distribution data for Mediterranean species (Legind, 2011) were downloaded from GBIF portal (http://www. gbif. org). The raw fossil pollen data (Leydet, 2011) is accessible for most used sites at the European Pollen database (http://www. europeanpollendatabase. net). For the remaining sites, d...
10.5194/cp-12-1137-2016
article
en
2,016
true
false
false
false
Earth and Planetary Sciences
https://openalex.org/fields/19
cc-by
The dissemination of textual personal information has become a key driver for innovation and value creation. However, due to the possible content of sensitive information, this data must be anonymized, which can reduce its usefulness for secondary uses. One of the most used techniques to anonymize data is generalizatio...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
As a consequence, the utility of the resulting datasets is improved without sacrificing the privacy goal. We also introduce a novel rating scale to classify the quality of the VGHs into categories to facilitate the interpretation of our quality metric for practitioners.
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
For example, individuals may suffer from discrimination or identity theft. Likewise, organizations may suffer from negative publicity, fines or other sanctions [18]. Hence, this data must be anonymized before being shared for analysis. Privacy-Preserving Data Publishing (PPDP) provides methods for publishing data with...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
The idea is that the original data loses its specificity, which reduces the probability of re-identifying the individuals contained in the published datasets. For example, one could generalize the terms "oncology clinic" and "cosmetic surgery hospital" to "medical institution" to protect the sensitive whereabouts of a ...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
Contributions. Considering these challenges, our research has centered on developing techniques to evaluate in an objective, quantifiable, and automatic way, the quality of textual data VGHs with the aim of improving their effectiveness for anonymizing data. In [5] we introduced the Generalization Semantic Loss metri...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
In this paper we extend our previous work [5] by proposing an enhanced VGH quality metric (d-GSL) which, based on a dynamic evaluation scheme, exploits the frequency distribution of the input datasets. In this manner, d-GSL can predict with higher accuracy the effectiveness that VGHs will have in anonymization. This ...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
II. BACKGROUND AND RELATED WORK The collection and dissemination of large amounts of personal information have allowed organizations to benefit from the exploitation of this data. Therefore, a lot of effort has been invested into studying PPDP techniques. In a typical PPDP scenario [4] , a trusted organization collect...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
This property is achieved by using VGHs. An example VGH is shown in Fig. 1. The leaves at level 0 (L0) represent the original distinct values of an attribute in the dataset. The ancestors at upper levels (L1 to L3) correspond to the candidate values used for the generalizations. The root node (at L3) corresponds to the...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
B. Static vs. Dynamic Selection of VGHs In the traditional anonymization process, the design and selection of VGHs are performed the first time that a dataset needs to be published by an organization. If a new dataset that belongs to the same domain (which has already been modeled) requires anonymization, the selected ...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
This is because the effectiveness of the VGHs (and the utility of the anonymized data) can be impacted by the characteristics of the input datasets. For this reason, our work proposes to use a dynamic strategy in the evaluation of the candidate VGHs. That is, VGHs should be evaluated depending on the input dataset, as ...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
C. Dynamic VGH Evaluation Scheme Overview In our work, the quality of the VGHs is represented by a score, which dynamically adjusts to the distribution of the input datasets. This strategy would allow a more accurate evaluation of the VGHs and thus, a better prediction of their effectiveness to conduct the data anonymi...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
Considering this, users need to have an efficient and effective manner of assessing candidate VGHs to decide which VGH best fits each input dataset. (4) Our proposed approach is a dynamic VGH evaluation scheme which captures, in a score denoted as d-GSL, the degree of data semantics that each VGH loses in their specifi...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
The semantic content of the ontologies is exploited by semantic similarity metrics to measure the proximity between the original values of a dataset and its possible generalizations; resembling human behavior regarding the judgment of similarity between two terms. (5) The output of our approach is one d-GSL score per ...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
D. Computing the Dynamic GSL Score In order to assess the quality of a VGH in terms of its semantics preservation and the input dataset, we evaluate the degree of information loss that could result from the transformations of the original values in the input dataset. To perform this, we first measure the semantic dista...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
4b). This is useful because under some schemes of anonymization, the generalization occurs at the domain level, that is, all the original values end at the same level of the VGH (e. g. , full-domain generalization [4] ). G(s 1 ,. , s n ) = ⎧ ⎪ ⎪ ⎨ ⎪ ⎪ ⎩ max i∈[i,n] s i , if G is maximum n i=1 s i n i=1 f i , if G is a...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
For example, the ancestor terms specified for "optometrist" are semantically incon- sistent. This is because although those terms refer to people involved in eye caring, "optometrist" is not an "eye-doctor" or "doctor". These type of issues can be easily detected by inspecting those transformations using our semanticba...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
These DS belong to the same domain but have different frequency distributions (as shown in the first column of Table III ): DS1 has a uniform distribution of 10% of each term; DS2 has 90% of "optometrist" terms (poorly-specified branch) and 10% divided among the rest; DS3 has 50% of "veterinarian" terms (well-specifie...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
For DS2, d-GSL showed an increment in the loss incurred by the VGH, as in this scenario 90% of the data falls into a branch where the losses are high. Finally for DS3, d-GSL showed a small increment as the data is distributed between well-defined and poorly-specified branches. Having obtained different d-GSL scores for...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
IV. EXPERIMENTAL SETUP Below, we present our experimental methodology and describe the testbed of VGHs, the datasets, and the evaluation criteria used in our experiments.
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
A. Experimental Methodology We conducted a series of experiments that pursued three objectives: (1) to investigate how the effectiveness of a VGH is subjective to the dataset that will employ it for anonymization; (2) to demonstrate how and why using a dynamic VGH quality assessment scheme, which considers the distribu...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
Later, we conducted the anonymization of datasets (see Section IV-B) using the candidate VGHs and the Datafly algorithm (a popular k-anonymity based algorithm) [21]. We tested different levels of privacy, varying the k-values ∈ [30. 100]. In this manner, all the anonymity levels of the candidate VGHs were covered, whi...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
B. Evaluated VGHs and Datasets Our testbed consisted of 252 VGHs. Those VGHs were created by perturbing the semantical content of an ideal VGH which was constructed by extracting the minimal taxonomy from WordNet (ontology widely used due to its broad coverage of concepts [14] ) for our evaluation data. To obtain a va...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
From this dataset, we focused on the attribute of occupation, which has the highest diversity of values. To test the generality of our solution, we derived multiple datasets per VGH. This strategy allowed us to considerably diversify the range of evaluated scenarios and to show the impact that a dataset can have over t...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
For example, for the dataset 70w20b10r, the 70% of the data was the worst branch, while 20% was the best, and 10% was distributed among the rest of the terms.
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
C. Evaluation Criteria VGH Quality. The quality of the VGHs is expressed in terms of our d-GSL metric and the s-GSL metric. As both metrics leverage on measures of semantic similarity, in our experiments we used two widely-used path-based metrics: Wu and Palmer (WUP) and Leacock and Chodorow (LCH) [14]. This strategy ...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
SSE measures the level of intra-group homogeneity in a group of anonymized records. SemILoss measures how semantically different the anonymized values are (on average) compared to the original ones. For both metrics, lower values are better, as it indicates a higher utility of the data. VGH Quality and Data Utility Cor...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
V. EXPERIMENTAL RESULTS Correlation Comparison. This analysis focused on assessing the capacity of the d-GSL and s-GSL metrics to capture (a priori) the effectiveness of the VGHs for anonymizing data. Figs. 3a and 3b show the correlation between the VGH quality metrics and the data utility per dataset. Fig. 3a shows...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
Results show that d-GSL continued to exhibit a stronger correlation (between 0. 79 and 0. 90) than s-GSL (between 0. 73 and 0. 85). To complement the validation, the highest and the lowest correlations obtained for each quality metric are shown in Figs. 4 and 5 (for the 50w40b10r and 05w90b05r dataset scenarios, resp...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
For the sake of brevity, we only present the comparison of ten sample VGHs (randomly picked from our testbed) that belong to the category 2 (according to their s-GSL score). Fig. 6a shows the d-GSL and s-GSL scores of the sample VGHs. Since s-GSL is computed using a static scheme, there is one s-GSL score per VGH. For...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
That is, the VGH that would yield the lower information loss in the anonymized data. These rankings are shown in Fig. 7a. For example, based on d-GSL and the 90w05b05r input dataset, the VGH3 would be the best. Whereas, if the input dataset were 05w90b05r, the best would be VGH2. In summary, Figs. 6a and 7a represent t...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
7a ) with the real rankings (shown in Fig. 7b ). Overall, the d-GSL metric predicted more accurately the performance of the VGHs than s-GSL. For example in Fig. 7a , s-GSL ranked the VGH2 in 2nd place, however its performance degraded with some datasets (e. g. , 70w20b10r and 90w05b05r) until falling to 9th place. In c...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
Instead, the best VGH can change depending on the input datasets. Hence, the utility of an anonymized dataset can be improved when the VGH evaluation is performed using a dynamic scheme (i. e. , d-GSL) instead of a static scheme (i. e. , s-GSL). VGH Quality Categories Comparison. This analysis assessed the empirical ca...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
It can also be noticed that the information losses are generally lower We also grouped the VGHs in categories based on their s-GSL scores; their behaviors were similar to the ones exhibited by the d-GSL categories (shown in Fig. 8 ). To validate which of the two category groups was more accurate, we compared the groups...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
Regarding the privacy implications of our approach, it does not affect the privacy goal set by users. In our experiments, the anonymized datasets improved their utility while still satisfying their corresponding kanonymity levels. Thus, the anonymized datasets kept the same protection and vulnerabilities of the privacy...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
VII. CONCLUSIONS AND FUTURE WORK In this paper, we proposed a dynamic VGH evaluation approach which exploits the frequency distribution of the input datasets. Our approach yields a score (d-GSL) which acts as predictor of the effectiveness of VGHs to perform the anonymization of data. The d-GSL score enables users to e...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
e. , d-GSL) instead of a static scheme (e. g. , s-GSL). Our results also demonstrated that d-GSL is more accurate than s-GSL, as it was better correlated with the utility of the datasets anonymized with the evaluated VGHs. Furthermore, we showed that VGHs classified in lower categories are more likely to yield a higher...
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
It would be also interesting to evaluate our solution further using more datasets and privacy models. Finally, we also intend to explore how to automatically generate well-defined VGHs or improve an "imperfect" VGH.
10.1109/iri.2016.13
article
en
2,016
true
false
false
false
Computer Science
https://openalex.org/fields/17
cc-by
ACKNOWLEDGMENTS This work was supported, in part, by Science Foundation Ireland grant 10/CE/I1855. We also thank A. Omar Portillo-Dominguez for the helpful discussions and reviews.
10.1109/iri.2016.13
article
en
2,016
false
false
true
false
Computer Science
https://openalex.org/fields/17
cc-by
Digital humanities scholars strongly need a corpus exploration method that provides topics easier to interpret than standard LDA topic models. To move towards this goal, here we propose a combination of two techniques, called Entity Linking and Labeled LDA. Our method identifies in an ontology a series of descriptive l...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
This will offer to digital humanities scholars an overview of the performance (and the drawbacks) of the different components of the pipeline, and to natural language processing researchers a series of baselines that will guide them towards improving each component of the method proposed. 4 hile the presented solutions...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
2014) and holds enormous potential for future political historians. The second dataset is the Enron Thread Corpus, which consists of e-mail threads from the Enron Corpus, a large database of over 600,000 emails generated by 158 employees of the Enron Corporation and acquired by the Federal Energy Regulatory Commission...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Dan Cohen (2006) , more than a decade ago, anticipated the issue that future political historians will encounter when considering the large abundance of sources 6 that public administration will leave us in the next decades. Our study intends to be a very first experimental attempt to deal with one of these new colle...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
One of the first papers that presents the task of topic labeling (Mei, Shen, and Zhai 2007) addresses the issue as an optimization problem involving minimizing the Kullback-Leibler divergence between word distributions, while maximizing the mutual information between a label and a topic model. A later approach (Lau ...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
This is an information extraction task that involves being able to recognise named entities in text 7 A completely different strategy (Chaney and Blei 2012) to improve the interpretability of topic model results relies on the use of data visualisation techniques. 8 See http://wiki. dbpedia. org/. 9 Some authors (Cha...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
In these tasks, humans have to detect a word or a topic which does not fit with the rest (Chang et al. 2009). Topic models can be evaluated per se, or their results can be evaluated against a gold standard. A way of doing it is to study the alignment between topic-results and classes in the dataset, or to use topic mo...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
e. , cherry-picked from the user based on data exploration) or automatically found on the basis of validation datasets or cross-validation techniques. We come back to the issue of selecting a value for k in Section 4. 3 paragraph "Label Selection and Ranking". 3. Entity-based Labeled LDA. In the final phase of our meth...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Datasets In this section we present the three datasets we adopted in our empirical evaluation. (1999) (2000) (2001) (2002) (2003) (2004) have been queried, via the SPARQL endpoint of the project 12. Each speech was considered as a single document. 13.
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
EuroParl EnronCorpus. The second dataset that was selected is the Enron Thread Corpus (Jamison and Gurevych 2013). This is based on the Enron dataset, a popular email dataset first presented by (Klimt and Yang 2004). However, unlike the original Enron dataset, the Enron Thread Corpus contains the emails grouped by em...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
(Nanni and Ruiz Fabo 2016) , we represented each party as a single document, containing all speeches by its politicians. The solution chosen for this article will help us understanding the differences in topics in a more fine-grained fashion. 14 https://www. ukp. tu-darmstadt. de/data/text-similarity/email-disentanglem...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Besides, the language in both datasets is sometimes relatively informal. On the other hand, the Clinton Email Dataset contains political topics and terms, and relates therefore to the very formal speeches that can be found in the EuroParl corpus. For these reasons, the third dataset is an interesting complement to the ...
10.4000/ijcol.392
article
en
2,016
true
true
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
ClintonCorpus By looking at the statistics on the three corpora (see Table 1 ), we notice that the vocabulary of the two email datasets is richer than EuroParl's vocabulary, as the email datasets are characterised by a large variety of informal expressions (e. g. "thx", "fyi", "tl:dr"). At the same time we can see tha...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Evaluation The evaluation we performed on our solution consists of three steps, aimed at assessing the reliability of each component in our study's pipeline. First of all, we conducted a general analysis on the use of entities as document labels. Next, we evaluated the quality of our ranking. Finally, we examined the r...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
The inter-annotator agreement on the label identification is κ = 0. 42, which thus confirms the difficulty and subjectivity of the task 20. As can be seen in Table 2 , in all three datasets annotators chose four labels on average. This possibly high number of labels often occurs since two highly related entities (e. g...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
The final number of labels selected per document highly depends on the needs of the researcher. In our case, based on the results of our study (presented in Table 3 ) and the behaviour of the annotators (see Table 2 ), we select labels by considering the first 4 entities in each document, ordered by tf-idf. However, ...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
e. choosing among topic-word sets given a label (Label Mode), or choosing among labels given one of topic-words (Term Mode), as explained when describing the design of the evaluation platform in Section 4. 2 above). In Table 4 we report the accuracy of user choices (i. e. the number of correct matches over the total ...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
e. terms that are found in Wikipedia but that are generally common nouns not clearly classified as a named entity). For instance, TagMe 2 achieves good results with both the IITB dataset (Kulkarni et al. 2009) , where most annotations are concepts (Waitelonis, Jürges, and Sack 2016, Table 1 ), and with the AIDA/CoNLL...
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
As mentioned above, while Entity Linking provides clear labels for the content of a document, it does not offer a direct notion of the proportion of the document that is related to each entity. Conversely, standard LDA's relevance scores do provide an estimate to what extent the topic is relevant for the document, but ...
10.4000/ijcol.392
article
en
2,016
false
true
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
The third data set is the Hillary Clinton Email Dataset. It consists of emails from Hillary Clinton's time as United States Secretary of State, during which.
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33
cc-by
Acknowledgments We want to thank all the people that used the first version of our evaluation platform at the 2016 Digital Humanities Conference in Krakow for providing us an initial gold standard and a series of useful feedbacks. Pablo Ruiz Fabo was supported through a PhD scholarship from Région Île-de-France.
10.4000/ijcol.392
article
en
2,016
false
false
true
false
Social Sciences
https://openalex.org/fields/33
cc-by
The third data set is the Hillary Clinton Email Dataset. It consists of emails from Hillary Clinton's time as United States Secretary of State, during which 11 http://www. talkofeurope. eu/ 12 http://linkedpolitics. ops. few. vu. nl/sparql/ 13 In order to get a general overview on the collection, in a previous work.
10.4000/ijcol.392
article
en
2,016
true
false
false
false
Social Sciences
https://openalex.org/fields/33