Professional Documents
Culture Documents
Download textbook Database Systems For Advanced Applications 22Nd International Conference Dasfaa 2017 Suzhou China March 27 30 2017 Proceedings Part I 1St Edition Selcuk Candan ebook all chapter pdf
Download textbook Database Systems For Advanced Applications 22Nd International Conference Dasfaa 2017 Suzhou China March 27 30 2017 Proceedings Part I 1St Edition Selcuk Candan ebook all chapter pdf
https://textbookfull.com/product/database-systems-for-advanced-
applications-25th-international-conference-dasfaa-2020-jeju-
south-korea-september-24-27-2020-proceedings-part-i-yunmook-nah/
https://textbookfull.com/product/database-systems-for-advanced-
applications-25th-international-conference-dasfaa-2020-jeju-
south-korea-september-24-27-2020-proceedings-part-ii-yunmook-nah/
https://textbookfull.com/product/database-systems-for-advanced-
applications-25th-international-conference-dasfaa-2020-jeju-
south-korea-september-24-27-2020-proceedings-part-iii-yunmook-
nah/
https://textbookfull.com/product/intelligent-information-and-
database-systems-9th-asian-conference-aciids-2017-kanazawa-japan-
april-3-5-2017-proceedings-part-i-1st-edition-ngoc-thanh-nguyen/
Database Systems for Advanced Applications DASFAA 2020
International Workshops BDMS SeCoP BDQM GDMA and AIDE
Jeju South Korea September 24 27 2020 Proceedings
Yunmook Nah
https://textbookfull.com/product/database-systems-for-advanced-
applications-dasfaa-2020-international-workshops-bdms-secop-bdqm-
gdma-and-aide-jeju-south-korea-september-24-27-2020-proceedings-
yunmook-nah/
https://textbookfull.com/product/business-information-
systems-20th-international-conference-bis-2017-poznan-poland-
june-28-30-2017-proceedings-1st-edition-witold-abramowicz-eds/
Selçuk Candan · Lei Chen
Torben Bach Pedersen · Lijun Chang
Wen Hua (Eds.)
LNCS 10177
Database Systems
for Advanced Applications
22nd International Conference, DASFAA 2017
Suzhou, China, March 27–30, 2017
Proceedings, Part I
123
Lecture Notes in Computer Science 10177
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen
Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Zurich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
C. Pandu Rangan
Indian Institute of Technology, Madras, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/7409
Selçuk Candan Lei Chen
•
Database Systems
for Advanced Applications
22nd International Conference, DASFAA 2017
Suzhou, China, March 27–30, 2017
Proceedings, Part I
123
Editors
Selçuk Candan Lijun Chang
Arizona State University University of New South Wales
Tempe - Phoenix, AZ Sydney, NSW
USA Australia
Lei Chen Wen Hua
Hong Kong University of Science The University of Queensland
and Technology Brisbane, QLD
Hong Kong Australia
China
Torben Bach Pedersen
Aalborg University
Aalborg
Denmark
LNCS Sublibrary: SL3 – Information Systems and Applications, incl. Internet/Web, and HCI
It is our great pleasure to welcome you to DASFAA 2017, the 22nd edition of the
International Conference on Database Systems for Advanced Applications (DASFAA),
which was held in Suzhou, China, during March 27–30, 2017.
The long history of Suzhou City has left behind many attractive scenic spots and
historical sites with beautiful and interesting legends. The elegant classical gardens, the
old-fashioned houses and delicate bridges hanging over flowing waters in the drizzling
rain, the beautiful lakes with undulating hills in lush green, and the exquisite arts and
crafts, among many other attractions, have made Suzhou a renowned historical and
cultural city full of eternal and poetic charm. Suzhou is best known for its gardens: the
Humble Administrator’s Garden, the Lingering Garden, the Surging Wave Pavilion,
and the Master of Nets Garden. These gardens weave together the best of traditional
Chinese architecture, painting, and arts. Suzhou is also known as the “Venice of the
East.” The city is sandwiched between Taihu Lake and Grand Canal. A network of
channels, criss-crossed with hump-backed bridges, give Suzhou an image of the city on
the water.
We were delighted to offer an exciting technical program, including three keynote
talks by Divesh Srivastava (AT&T Research), Christian S. Jensen (Aalborg Univer-
sity), and Victor Chang (Xi’an Jiaotong University and Liverpool University), one
10-year best paper award presentation; a demo session with four demonstrations; two
industry sessions with nine paper presentations; three tutorial sessions; and of course a
superb set of research papers. This year, we received 300 submissions to the research
track, each of which went through a rigorous review process. Specifically, each paper
was reviewed by at least three Program Committee (PC) members, followed by a
discussion led by the PC co-chairs. Several papers went through a shepherding process.
Finally, DASFA 2017 accepted 73 full research papers, yielding an acceptance ratio of
24.3%.
Four workshops were selected by the workshop co-chairs to be held in conjunction
with DASFAA 2017. They are the 4th International Workshop on Big Data Man-
agement and Service (BDMS 2017), the Second Workshop on Big Data Quality
Management (BDQM 2017), the 4th International Workshop on Semantic Computing
and Personalization (SeCoP), and the First International Workshop on Data Manage-
ment and Mining on MOOCs (DMMOOC 2017).
The workshop papers are included in a separate volume of the proceedings also
published by Springer in its Lecture Notes in Computer Science series.
The conference received generous financial support from Soochow University. The
conference organizers also received extensive help and logistic support from the
DASFAA Steering Committee and the Conference Management Toolkit Support Team
at Microsoft.
VI Preface
We are grateful to the general chairs, Karl Aberer, EPFL, Switzerland, Peter
Scheuermann, Northwestern University, USA, and Kai Zheng, Soochow University,
China, the members of the Organizing Committee, and many volunteers, for their great
support in the conference organization. Special thanks also go to the DASFAA 2017
local Organizing Committee: Zhixu Li and Jiajie Xu, both from Soochow University,
China. Finally, we would like to take this opportunity to thank the authors who sub-
mitted their papers to the conference and the PC members and external reviewers for
their expertise and help in evaluating the submissions.
General Co-chairs
Karl Aberer EPFL, Switzerland
Peter Scheuermann Northwestern University, USA
Kai Zheng Soochow University, China
Workshops Co-chairs
Zhifeng Bao RMIT, Australia
Goce Trajcevski Northwestern University, USA
Tutorial Co-chairs
Katja Hose Aalborg University, Denmark
Huiping Cao New Mexico State University, USA
Panel Chair
Xuemin Lin University of New South Wales, Australia
Proceedings Co-chairs
Lijun Chang University of New South Wales, Australia
Wen Hua University of Queensland, Australia
VIII Organization
Publicity Co-chairs
Bin Yang Aalborg University, Denmark
Xiang Lian University of Texas-Pan American, USA
Maria Luisa Sapino University of Turin, Italy
Conference Secretary
Yan Zhao Soochow University, China
Webmaster
Yang Li Soochow University, China
Program Committee
Amr El Abbadi University of California at Santa Barbara, USA
Alberto Abello UPC Barcelona, Spain
Divyakant Agrawal University of California at Santa Barbara, USA
Marco Aldinucci University of Turin, Italy
Ira Assent Aarhus University, Denmark
Rafael Berlanga Llavori University Jaume I, Spain
Francescho Bonchi ISI Foundation, Italy
Selcuk Candan Arizona State University, USA
Huiping Cao New Mexico State University, USA
Barbara Catania Università di Genova, Italy
Qun Chen Northwestern Polytechnical University, China
Reynold Cheng University of Hong Kong, SAR China
Wonik Choi Inha University, South Korea
Gao Cong Nanyang Technological University, Singapore
Bin Cui Peiking University, China
Lars Dannecker SAP, Germany
Hasan Davulcu Arizona State University, USA
Ugur Demiryurek University of Southern California, USA
Francesco Di Mauro University of Turin, Italy
Curtis Dyreson Utah State University, USA
Hakan Ferhatosmanoglu Bilkent University, Turkey
Organization IX
Additional Reviewers
Jinpeng Chen Beihang University, China
Xilun Chen Arizona State University, USA
Yu Cheng Turn Inc., USA
Alexander Crosdale UC Merced, USA
Tiziano De Matteis University of Pisa, Italy
Vasilis Efthymiou University of Crete, Greece
Roberto Esposito University of Turin, Italy
Yixiang Fang The University of Hong Kong, SAR China
Christian Frey LMU Munich, Germany
Yash Garg Arizona State University, USA
Concorde Habineza George Mason University, USA
Jiafeng Hu The University of Hong Kong, SAR China
Zhiyi Huang UC Merced, USA
Zhipeng Huang The University of Hong Kong, SAR China
Shengyu Huang Arizona State University, USA
XII Organization
Network Embedding
Data Mining
Text Mining
Supervised Intensive Topic Models for Emotion Detection over Short Text . . . . 408
Yanghui Rao, Jianhui Pang, Haoran Xie, An Liu, Tak-Lam Wong,
Qing Li, and Fu Lee Wang
Recommendation
Tutorials
Spatial Databases
Efficiently Mining High Utility Co-location Patterns from Spatial Data Sets
with Instance-Specific Utilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
Lizhen Wang, Wanguo Jiang, Hongmei Chen, and Yuan Fang
Demos
1 Introduction
In recent years, producing large-scale datasets has been vital for both research
and industry. Traditional strategies, mostly hiring motivated experts, are not
competent enough when collecting datasets like ImageNet [1] and TinyImages
[2] which contain millions of samples. On the other hand, crowdsourcing services,
like AMT (Amazon Mechanical Turk) [3] or CrowdFlower [4], offer a convenient
way to collect datasets by distributing tasks to workers (sources) all over the
world.
For its efficiency and effectiveness in producing large-scale datasets, crowd-
sourcing has become increasingly popular. As sources are not necessarily experts,
tasks are often distributed more than once, which results in the noisy and con-
flicted answers. Therefore, how to aggregate answers and infer the ground truth
has been a critical challenge in crowdsourcing.
c Springer International Publishing AG 2017
S. Candan et al. (Eds.): DASFAA 2017, Part I, LNCS 10177, pp. 3–18, 2017.
DOI: 10.1007/978-3-319-55753-3 1
4 Y. Du et al.
tasks with the task-specific description but are not competent for those with
non-specific description. Other approaches, which model fine-grained reliability
for different answers, limit the tasks to those whose candidate answers are non-
specific. As consequences, (1) none of these approaches can deal with the Writing
tasks or the Transcription tasks, and (2) they can only utilize a subset of the
global dataset to help infer the ground truth. Therefore, it is crucial to design a
general fine-grained truth discovery approach.
To tackle with challenges above, we consider the clusters in both tasks and
sources. It has been observed that members of the same task cluster or source
cluster share identical patterns. For instance, the pattern shared in a source
cluster is sources’ rugged reliability levels for different task clusters. More specif-
ically, sources can have different reliability levels for various task clusters, but
for a particular task cluster, sources coming from the same cluster show an equal
reliability level. The patterns of task clusters are similar to that of source clusters.
Therefore, we can first learn the patterns of task clusters and source clusters,
then derive correct answers for target tasks based on the learned patterns.
In this paper, we propose a general fine-grained truth discovery approach
to aggregate crowdsourced answers. The proposed method contains two integral
components: kl-means and Pattern-based Truth Discovery (PTD). With the aid
of standard tasks whose ground truth are known, kl-means directly employs a
co-clustering reliability model on the correctness matrix to learn the patterns
of sources and tasks simultaneously. Then PTD further uses captured patterns
to help infer the ground truth. Since these two components do not have any
additional requirements for tasks, like the task-specific description or non-specific
answers, our approach is compatible with all tasks. To the best of our knowledge,
this is the first work to propose a general fine-grained method with consideration
of the patterns of task clusters and source clusters. One important feature of our
approach is the utilization of co-clustering reliability model and standard tasks,
which allows us to learn the cluster patterns directly from the correctness matrix.
Therefore, our method can better demonstrate the relationship between tasks
and sources and produce more accurate inference.
In summary, three main contributions of this work are as follows:
(1) To the best of our knowledge, we are the first to recognize the compatibility
issues of state-of-art fine-grained truth discovery methods and propose a
general one to aggregate noisy crowdsourced answers.
(2) We employ a co-clustering reliability model to learn the patterns of task
clusters and source clusters simultaneously and use captured patterns to
derive a more accurate inference of ground truth.
(3) We empirically show that the proposed method outperforms existing general
truth discovery methods.
2 Problem Formulation
In this section, we first introduce some basic notations used in this paper, then
give a formal definition of our problem.
6 Y. Du et al.
The inputs of our method are m standard tasks T = {ti }m 1 , m target tasks
m+m m+m ,n
T = {ti }m+1 , n sources U = {uj }1 , answers matrix A = {aij }i=1,j=1 and cor-
n
Definition 2. Answer element aij denotes the answer given by source uj to task
ti . aij = b, 1 ≤ b ≤ B means source uj select task ti ’s candidate answer cib as
his answer, aij = 0 means source uj has not answered task ti .
Definition 4. Estimated ground truth ei for task ti is the most accurate answer
provided by sources.
Based on these definitions, we can formally define our problem as: given m
m+m
standard tasks T = {ti }m
1 , m target tasks T = {ti }m+1 , n sources U = {uj }1 ,
n
m+m ,n
answers matrix A = {aij }i=1,q=1 and correctness matrix R = {rij }m,n i=1,q=1 , our
goals are learning k task clusters I = {Ip }1 , l source clusters {Jq }1 , co-clustering
k l
Therefore, we can divide tasks into k clusters, sources into l clusters by the
following functions:
ρ : {t1 , t2 · · · tm } → {1, 2 · · · k}
(1)
γ : {u1 , u2 · · · un } → {1, 2 · · · l},
where ρ(ti ) = r implies that task ti is in task cluster r, γ(uj ) = s implies that
source uj is in source cluster s.
Let I = {Ip }k1 denote the k task clusters and J = {Jq }l1 denote the source
clusters. A co-cluster is referred as the submatrix of correctness matrix R deter-
mined by a task cluster Ip and a source cluster Jq .
We evaluate the homogeneity of co-cluster by the sum of squared differences
between each entry and the mean of this co-cluster which was used by Hartigan
et al. [15]. Consider that the correctness matrix could be highly sparse since it
is unnecessary for sources to perform all tasks. We compute the residue of the
element rij differently based on whether source uj has answered task ti . Let
value −1 represent the unfilled items in the correctness matrix. Formally:
rij − ηpq , if rij = −1
hij = (2)
0, otherwise.
When source uj has answered task ti , the residue of rij is hij = rij − ηpq ,
where p is the cluster label of task ti , q is the cluster label of source uj , ηpq is
the mean of submatrix determined by Ip and Jq . But, when source uj has not
answered task ti , we set hij to 0, which in fact ignores the loss caused by unfilled
elements.
The objective of this co-clustering reliability model is to learn the patterns
of both tasks and sources. Hence, the objective function is to minimize the total
squared residues of all co-clusters, which is shown as follows.
L= Lpq = h2ij , (3)
Ip ,Jq Ip ,Jq i∈Ip ,j∈Jq
where Ip , Jq enumerates all co-clusters, Lpq represents the sum squared residues
of co-cluster determined by Ip and Jq .
4 Algorithms
The proposed approach estimates the ground truth by employing two integral
components: (1) using kl-means to learn the co-clustering reliability model from
sources performance on standard tasks, which contains the patterns of clusters,
and (2) infer the ground truth of target tasks with the help of capture patterns.
4.1 kl-means
We propose kl-means, an extended version of the k-means algorithm, to learn
co-clustering reliability of tasks and sources. Differing from previous work
Another random document with
no related content on Scribd:
⅓ cup soft shortening
1 cup brown sugar
1½ cups black molasses
Stir in ...
Stir in ...
Chill dough. Roll out very thick (½″). Cut with 2½″ round cutter. Place
far apart on lightly greased baking sheet. Bake until, when touched
lightly with finger, no imprint remains.
temperature: 350° (mod. oven).
time: Bake 15 to 18 min.
amount: 2⅔ doz. fat, puffy 2½″ cookies.
FROSTED GINGIES
Follow recipe above—and frost when cool with Simple White Icing
(recipe below).
GINGERBREAD BOYS
Make holidays gayer than ever.
Follow recipe above—and mix in 1 more cup sifted GOLD MEDAL
Flour. Chill dough. Roll out very thick (½″). Grease cardboard
gingerbread boy pattern, place on the dough, and cut around it with
a sharp knife. Or use a gingerbread boy cutter. With a pancake
turner, carefully transfer gingerbread boys to lightly greased baking
sheet. Press raisins into dough for eyes, nose, mouth, and shoe and
cuff buttons. Use bits of candied cherries or red gum drops for coat
buttons; strips of citron for tie. Bake. Cool slightly, then carefully
remove from baking sheet. With white icing, make outlines for collar,
cuffs, belt, and shoes.
amount: About 12 Gingerbread Boys.
1 cup molasses
½ cup shortening
1 tsp. soda
Chill dough. Roll out very thin (¹⁄₁₆″). Cut into desired shapes. Bake
until, when touched lightly, no imprint remains. (Over-baking gives a
bitter taste.)
temperature: 350° (mod. oven).
time: Bake 5 to 7 min.
amount: About 6 doz. 2½″ cookies.
Chill dough. Roll out thick (¼″). Cut into desired shapes. Place 1″
apart on lightly greased baking sheet. Bake until, when touched
lightly with finger, no imprint remains. When cool, ice and decorate
as desired.
temperature: 375° (quick mod. oven).
time: Bake 8 to 10 min.
amount: About 5 doz. 2½″ cookies.
LIGHT DOUGH
For bells, stockings, stars, wreaths, etc.
Follow recipe for Dark Dough above except substitute honey for
molasses, and granulated sugar for brown. Use 1 tsp. vanilla in
place of cinnamon and ginger.
TO DECORATE
Use recipe for Decorating Icing (p. 31) (thin the icing for spreading).
For decorating ideas, see picture on preceding page. Sugar in
coarse granules for decorating is available at bakery supply houses.
STARS
Cover with white icing. Sprinkle with sky blue sugar.
WREATHS
Cut with scalloped cutter ... using smaller
cutter for center. Cover with white icing.
Sprinkle with green sugar and decorate with clusters
of berries made of red icing—leaves of green icing—
to give the realistic effect of holly wreaths.
BELLS
Outline with red icing. Make clapper of red icing. (A
favorite with children.)
STOCKINGS
Sprinkle colored sugar on toes and heels before
baking. Or mark heels and toes of baked cookies
with icing of some contrasting color.
CHRISTMAS TREES
Spread with white icing ... then sprinkle with green sugar.
Decorate with silver dragées and tiny colored candies.
TOYS
(Drum, car, jack-in-the-box, etc.):
Outline shapes with white or colored icing.
ANIMALS
(Reindeer, camel, dog, kitten, etc.): Pipe icing
on animals to give effect of bridles, blankets,
etc.
BOYS AND GIRLS
Pipe figures with an icing to give desired effects:
eyes, noses, buttons, etc.
½ cup honey
½ cup molasses
Cool thoroughly
Stir in ...
Mix in ...
GLAZING ICING
Boil together 1 cup sugar and ½ cup water until first indication of a
thread appears (230°). Remove from heat. Stir in ¼ cup
confectioners’ sugar and brush hot icing thinly over cookies. (When
icing gets sugary, reheat slightly, adding a little water until clear
again.)
★ NURNBERGER
Round, light-colored honey cakes from the famed old City of Toys.
Follow recipe above—except in place of honey and molasses use
1 cup honey; and reduce spices (using ¼ tsp. cloves, ½ tsp. allspice,
and ½ tsp. nutmeg ... with 1 tsp. cinnamon).
Roll out the chilled dough ¼″ thick. Cut into 2″ rounds. Place on
greased baking sheet. With fingers, round up cookies a bit toward
center. Press in blanched almond halves around the edge like petals
of a daisy. Use a round piece of citron for each center. Bake just until
set. Immediately brush with Glazing Icing (above). Remove from
baking sheet. Cool, and store to mellow.
amount: About 6 doz. 2½″ cookies.
TO “MELLOW” COOKIES
... store in an air-tight container for a few days. Add a cut orange or apple; but fruit
molds, so change it frequently.
ZUCKER HÜTCHEN (Little Sugar Hats)
From the collection of Christmas recipes by the Kohler Woman’s Club of Kohler,
Wisconsin.
Mix together thoroughly ...
6 tbsp. soft butter
½ cup sugar
1 egg yolk
Stir in ...
2 tbsp. milk
Mix in ...
Chill dough. Roll thin (⅛″). Cut into 2″ rounds. Heap 1 tsp. Meringue
Frosting (recipe below) in center of each round to make it look like
the crown of a hat. Place 1″ apart on greased baking sheet. Bake
until delicately browned.
temperature: 350° (mod. oven).
time: Bake 10 to 12 min.
amount: About 4 doz. 2″ cookies.
MERINGUE FROSTING
Beat 1 egg white until frothy. Beat in gradually 1½ cups sifted
confectioners’ sugar and beat until frosting holds its shape. Stir in ½
cup finely chopped blanched almonds.
SCOTCH SHORTBREAD
Old-time delicacy from Scotland ... crisp, thick, buttery.
Mix together thoroughly
...
Stir in ...
Mix thoroughly with hands. Chill dough. Roll out ⅓ to ½″ thick. Cut
into fancy shapes (small leaves, ovals, squares, etc.). Flute edges if
desired by pinching between fingers as for pie crust. Place on
ungreased baking sheet. Bake. (The tops do not brown during
baking ... nor does shape of the cookies change.)
temperature: 300° (slow oven).
time: Bake 20 to 25 min.
amount: About 2 doz. 1″ × 1½″ cookies.
Stir in ...
Stir in ...
*In place of the almonds, you may use 1 tsp. vanilla flavoring and 1
tsp. almond flavoring.
Chill dough. Press dough into Sandbakels molds (or tiny fluted tart
forms) to coat inside. Place on ungreased baking sheet. Bake until
very delicately browned. Tap molds on table to loosen cookies and
turn them out of the molds.
temperature: 350° (mod. oven).
time: Bake 12 to 15 min.
amount: About 3 doz. cookies.
MOLDED COOKIES Mold ’em fast with
a fork or glass!
1 With hands, roll dough 2 Flatten balls of dough 3 Cut pencil-thick strips
into balls or into long, with bottom of a glass ... and shape as directed
pencil-thick rolls, as dipped in flour (or with a ... as for Almond
indicated in recipe. damp cloth around it), or Crescents (p. 41) or
with a fork—crisscross. Berliner Kranser (p. 42).
DATE-OATMEAL COOKIES
Mix together thoroughly ...
Stir in ...
2 cups rolled oats
1½ cups cut-up dates
¾ cup chopped nuts
Chill dough. Roll into balls size of large walnuts. Place 3″ apart on
lightly greased baking sheet. Flatten (to ¼″) with bottom of glass
dipped in flour. Bake until lightly browned.
temperature: 375° (quick mod. oven).
time: Bake 10 to 12 min.
amount: About 4 doz. 2½″ cookies.
Chill dough. Roll into balls size of large walnuts. Place 3″ apart on
lightly greased baking sheet. Flatten with fork dipped in flour ...
crisscross. Bake until set ... but not hard.
temperature: 375° (quick mod. oven).
time: Bake 10 to 12 min.
amount: About 3 doz. 2½″ cookies.
Roll into 1″ balls. Dip in slightly beaten egg whites. Roll in finely
chopped nuts (¾ cup). Place about 1″ apart on ungreased baking
sheet. Bake 5 min. Remove from oven. Quickly press thumb gently
on top of each cooky. Return to oven and bake 8 min. longer. Cool.
Place in thumbprints a bit of chopped candied fruit, sparkling jelly, or
tinted confectioners’ sugar icing.
temperature: 375° (quick mod. oven).
time: Bake 5 min., then 8 min.
amount: About 2 doz. 1½″ cookies.
Mix in ...
Chill dough. Roll into balls the size of walnuts. Dip tops in slightly
beaten egg white, then sugar. Place sugared-side-up 2″ apart on
ungreased baking sheet. Bake until delicately browned. The balls
flatten some in baking and become glazed.
temperature: 400° (mod. hot oven).
time: Bake 12 to 15 min.
amount: About 3 doz. 1½″ cookies.
ALMOND CRESCENTS
Richly delicate, buttery. Party favorites.
Mix together thoroughly ...
Chill dough. Roll with hands pencil-thick. Cut in 2½″ lengths. Form
into crescents on ungreased baking sheet. Bake until set ... not
brown. Cool on pan. While slightly warm, carefully dip in 1 cup
confectioners’ sugar and 1 tsp. cinnamon mixed.
temperature: 325° (slow mod. oven).
time: Bake 14 to 16 min.
amount: About 5 doz. 2½″ cookies.
LEMON SNOWDROPS
Refreshing, lemony ... with snowy icing.
Follow recipe for English Tea Cakes above—except use 2 tbsp.
lemon juice and 1 tbsp. water in place of the milk. Add 2 tsp. grated
lemon rind. Omit citron and currants. Mix in ½ cup chopped nuts.
Chill dough. Roll into balls and bake. Then roll in confectioners’
sugar.
BUTTER FINGERS
Nut-flavored, rich buttery party cookies.
Follow recipe for Almond Crescents—except in place of almonds use
black walnuts or other nuts, chopped. Cut into finger lengths and
bake. While still warm, roll in confectioners’ sugar. Cool, and roll in
the sugar again.
Festive cookies for the holidays ... ideal for Christmas boxes.