trigram

This is a table of type trigram and their frequencies. Use it to search & browse the list to learn more about your study carrel.

trigram	frequency
of machine learning	37
machine learning and	29
the use of	29
a machine learning	25
a set of	25
in order to	24
comment by daniel	24
by daniel johnson	24
as well as	23
be able to	19
and machine learning	16
the number of	15
library of congress	15
generative adversarial networks	14
proceedings of the	14
in machine learning	14
learning and deep	13
one of the	13
and deep learning	13
the machine learning	13
machine learning is	13
an site ehost	12
direct true db	12
the process of	12
the ability to	11
the creation of	11
machine learning in	11
the full text	11
the idea of	11
in the case	11
the case of	11
generative machine learning	11
by mark dehmlow	10
in proceedings of	10
of artificial intelligence	10
machine learning techniques	10
machine learning algorithms	10
comment by mark	10
an ai system	10
archives and libraries	10
all of the	10
at the same	9
are likely to	9
learning is a	9
chicago place names	9
given a set	9
such as a	9
in this case	9
in the library	9
some of the	9
there is no	9
large amount of	8
full text of	8
reading chicago reading	8
were able to	8
disciplinary ml research	8
the new york	8
natural language processing	8
on the other	8
as a result	8
the training data	8
the library of	8
the trolley problem	8
the context of	8
there are many	8
the university of	8
it is also	8
new york times	8
the same time	8
the other hand	8
there is a	8
if you are	7
it is not	7
to be able	7
different types of	7
is likely to	7
that can be	7
international conference on	7
the question of	7
machine learning to	7
is important to	7
may not be	7
collections as data	7
and it is	7
amount of data	7
bth an site	7
it would be	7
of generative machine	7
chicago place name	7
and the library	7
we were able	7
db bth an	7
creation of a	7
of the machine	7
it is important	7
true db bth	7
the results of	7
a chicago place	7
a place name	7
automated information environment	7
you want to	7
it can be	7
allow us to	7
part of the	7
a large amount	7
the development of	7
computer vision and	6
in other words	6
of the association	6
to create a	6
learning and the	6
to work with	6
the fact that	6
need to be	6
the field of	6
with machine learning	6
by machine learning	6
whether or not	6
in the end	6
can be used	6
and artificial intelligence	6
parts of the	6
can also be	6
of the process	6
we do not	6
based on the	6
all of these	6
plain text files	6
oklahoma state university	6
machine learning as	6
in such a	6
the beginning of	6
of generative adversarial	6
of data and	6
the digital humanities	6
this is a	6
intelligence and machine	6
the latent space	6
such as the	6
back to the	6
topic modeling tool	6
a list of	6
it does not	5
deep learning applications	5
learning in the	5
it comes to	5
training and testing	5
the results to	5
level of autonomy	5
this can be	5
to the nearest	5
impact on the	5
to build a	5
be used to	5
an ai algorithm	5
the names of	5
and use the	5
of the ieee	5
of autonomy and	5
and ethical sensitivity	5
test data should	5
can be a	5
a lot of	5
to machine learning	5
english and italian	5
of an ai	5
of deep learning	5
in a new	5
the content of	5
book one chicago	5
for a machine	5
the lack of	5
to learn how	5
as long as	5
autonomous ai systems	5
learning as a	5
of the library	5
learning techniques to	5
that are not	5
we will be	5
in this chapter	5
the quality of	5
in this way	5
the result of	5
is one of	5
that you can	5
lxh an site	5
at the university	5
ways in which	5
at all levels	5
journal of the	5
it is the	5
in the process	5
and at the	5
db lxh an	5
related to the	5
is used to	5
machine learning applications	5
springer international publishing	5
learning can be	5
autonomy and ethical	5
a group of	5
a series of	5
of the project	5
intelligent as a	5
conference on computer	5
to focus on	5
many of the	5
will help you	5
machine learning solution	5
the value of	5
true db lxh	5
for information science	5
is a good	5
when it comes	5
as a human	5
a number of	5
do not have	5
the association for	5
compared to the	5
library technology reports	5
on computer vision	5
in the past	5
powered military robots	5
so that the	5
researchers at all	4
to generate new	4
to each other	4
fields of study	4
use the results	4
machine learning research	4
using machine learning	4
you have a	4
the nature of	4
neural information processing	4
use machine learning	4
and data mining	4
it learns to	4
data that is	4
of the most	4
very close to	4
be difficult to	4
good balance of	4
for machine learning	4
use of its	4
use of the	4
powered automated information	4
not have to	4
new ways to	4
types of data	4
for computational linguistics	4
as the data	4
chicago reading project	4
and pattern recognition	4
this is called	4
of machine morality	4
given the full	4
library collections and	4
strengths and weaknesses	4
culture of innovation	4
a markov chain	4
our moral intuition	4
of the american	4
will need to	4
of the nd	4
along the way	4
in this essay	4
but they are	4
role in the	4
balance of time	4
able to obtain	4
the final results	4
make use of	4
of a machine	4
to compare the	4
the third coast	4
we aim to	4
it should be	4
amounts of data	4
which is a	4
learning in libraries	4
in the three	4
artificial intelligence in	4
learning and artificial	4
on the final	4
set of previously	4
but it can	4
to develop a	4
the concept of	4
that in the	4
vision and pattern	4
your research question	4
learning and ai	4
of its digital	4
of the city	4
pine mountain settlement	4
in the digital	4
will be able	4
for the humanities	4
it can also	4
this is the	4
the next step	4
what it is	4
large number of	4
if you can	4
intellectual isolation and	4
the language of	4
of a dataset	4
as they are	4
the pine mountain	4
the contentdm instance	4
of time alone	4
the advent of	4
we tried to	4
in the literature	4
likely to be	4
machine learning tools	4
of the scholarly	4
the model to	4
as we have	4
the scholarly communications	4
this type of	4
is a very	4
especially if you	4
column is a	4
ieee transactions on	4
are not as	4
there are two	4
goodfellow et al	4
research and scholarship	4
the potential to	4
a way that	4
machine learning are	4
place name recognizer	4
the scope of	4
to be a	4
are able to	4
uses machine learning	4
of plain text	4
that will help	4
between the two	4
involved in the	4
data and the	4
continue to be	4
the performance of	4
supervised and unsupervised	4
some of them	4
to make the	4
an array of	4
historical social network	4
mit technology review	4
much of the	4
information processing systems	4
to identify the	4
to the library	4
the people who	4
through the use	4
top strengths of	4
in the world	4
of a chicago	4
the top strengths	4
in support of	4
automated information systems	4
collections and services	4
well as the	4
in the future	4
functionality of computers	4
machine learning process	4
as much as	4
in the archive	4
to do with	4
machine learning can	4
association for computational	4
understanding of the	4
more and more	4
machine learning systems	4
may need to	4
we have seen	4
we wanted to	4
quantitative and qualitative	4
the data that	4
mountain settlement school	4
close to each	4
in the same	3
the capacity for	3
because we have	3
in the photographs	3
libraries and librarians	3
of collaborating with	3
each step of	3
the american society	3
machine learning tasks	3
the diversity of	3
their ability to	3
of the internet	3
social network of	3
cohen and mario	3
one book one	3
organisms that could	3
to have a	3
we could not	3
in which the	3
information about the	3
be aware of	3
commanders and soldiers	3
machines do not	3
topic modeling to	3
able to identify	3
significant ethical challenges	3
faces of named	3
and use of	3
artificial intelligence and	3
adventures of augie	3
the data and	3
it into a	3
a new training	3
a machine with	3
teaching and learning	3
refers to the	3
based on a	3
an application of	3
the lower mississippi	3
be reflected in	3
according to the	3
of an individual	3
isolation and bigotry	3
services and operations	3
in all areas	3
two neural networks	3
learns to produce	3
to make decisions	3
to maximize the	3
of the data	3
the time and	3
the most effective	3
oxford university press	3
extracted from other	3
the role of	3
included in the	3
science and engineering	3
immutable data storage	3
also need to	3
we had a	3
in eastern kentucky	3
will only be	3
needs to be	3
the learning process	3
you need to	3
the power of	3
to research and	3
and in the	3
a good balance	3
for libraries to	3
to train a	3
we have the	3
do the work	3
save you from	3
full moral agency	3
it is difficult	3
people in the	3
matrix of vectors	3
the purpose of	3
military commanders and	3
the contents of	3
that may be	3
the ethical and	3
how can librarians	3
the library to	3
annual meeting of	3
can be difficult	3
the efficiency of	3
the relationships between	3
as a way	3
the most recent	3
a computer scientist	3
areas of the	3
source of labeled	3
its digital collection	3
the world of	3
the age of	3
there will be	3
step in the	3
the phenomenon of	3
defined as the	3
the reading chicago	3
chicago public library	3
the date and	3
handle their differences	3
a data challenge	3
the problem of	3
improved to better	3
can then be	3
the results are	3
a couple of	3
a classification problem	3
the strengths and	3
in classical chinese	3
a collection of	3
libraries and archives	3
to learn about	3
it is generally	3
in the comments	3
on the idea	3
use of machine	3
the program has	3
the national endowment	3
have to use	3
with our partners	3
even if you	3
and they can	3
the public and	3
report on the	3
you have to	3
machine learning workflow	3
training data and	3
in the development	3
would be of	3
effective as the	3
to produce a	3
a gan that	3
it may be	3
what level of	3
in a specific	3
the way we	3
ai systems are	3
use of pmss	3
which can be	3
the evolution of	3
from the same	3
as graph theory	3
an algorithm that	3
their problems well	3
mathematical subject classification	3
to think about	3
the warmth of	3
is that the	3
the library as	3
utilitarianism and deontology	3
early chinese empires	3
but the resulting	3
it will be	3
the list of	3
solicitation for a	3
we needed to	3
the work of	3
a way to	3
meeting of the	3
of natural language	3
machine learning project	3
be possible to	3
deep learning techniques	3
a culture of	3
in the research	3
their differences creatively	3
may also be	3
names extracted from	3
of how the	3
jus in bello	3
only be as	3
you will want	3
as a consequence	3
place name is	3
the most important	3
one place to	3
generate metadata for	3
highly technical articles	3
a file name	3
as we do	3
of sentences that	3
associated with the	3
such a way	3
the structure of	3
will be to	3
can serve as	3
material in a	3
if you wish	3
markov chains trained	3
the three oboc	3
there is also	3
sentences that mention	3
level of machine	3
learning is an	3
google flu trends	3
machine learning book	3
to determine the	3
jason cohen and	3
to find the	3
go to step	3
machine learning system	3
feel very close	3
allows you to	3
as we suggest	3
in light of	3
ai system is	3
cultural heritage institution	3
question of whether	3
the test data	3
so that you	3
of the region	3
tit per job	3
to create new	3
of tools and	3
these types of	3
the journal of	3
the intersection of	3
place name dataset	3
and how they	3
long as we	3
thing to do	3
have not been	3
the amount of	3
ai algorithm to	3
most of the	3
the time of	3
generate new data	3
as a source	3
do not learn	3
per job title	3
of research and	3
ought to be	3
access to the	3
the topic of	3
and discovery of	3
machine learning program	3
has been trained	3
a leap forward	3
the accuracy of	3
the sentiment of	3
machine learning with	3
in addition to	3
in the way	3
found that the	3
the library could	3
on the right	3
in the text	3
large data sets	3
the difference between	3
file path for	3
the adventures of	3
labeled training data	3
of the th	3
that machine learning	3
program has been	3
of the two	3
the ways in	3
create a model	3
discuss their problems	3
the real data	3
the plain text	3
as effective as	3
the data from	3
capacity to sustain	3
computers in libraries	3
what kind of	3
deep learning pilot	3
scope of the	3
of digital scholarship	3
it is possible	3
in a machine	3
american society for	3
com ericleasemorgan bringing	3
is also a	3
the generator learns	3
as in a	3
the importance of	3
an action is	3
if you want	3
national endowment for	3
with other cultural	3
names in the	3
against the plain	3
you wish to	3
a source of	3
of augie march	3
will likely be	3
in your data	3
the social network	3
a focus on	3
change in the	3
ai and machine	3
by the use	3
as a new	3
a model to	3
as a library	3
can be applied	3
is difficult to	3
as a relational	3
gaps in the	3
traditional machine learning	3
you will be	3
to look at	3
the challenges in	3
of a cat	3
step of the	3
to increase the	3
of a gan	3
of highly technical	3
set of plain	3
still need to	3
in neural information	3
marc is a	3
we have also	3
be as effective	3
data to the	3
and the future	3
to test the	3
is morally wrong	3
is still a	3
from the first	3
characteristics of the	3
place names extracted	3
discovery and use	3
and mario nakazawa	3
it is to	3
of the ai	3
machine learning has	3
information science technology	3
job title tit	3
the library or	3
and the people	3
set of data	3
it is a	3
applications for libraries	3
nature of the	3
to leverage the	3
to find a	3
positive and negative	3
we discovered that	3
raises the question	3
endowment for the	3
the goal of	3
despite the fact	3
presented at the	3
only as good	3
learn how to	3
at the time	3
the complexity of	3
representative of the	3
a part of	3
would not have	3
if it is	3
to your data	3
a library service	3
that it is	3
machine learning pipeline	3
types of projects	3
a good example	3
the top level	3
an image of	3
they are trained	3
institutional memory loss	3
library services and	3
in a researchers	3
a controlled vocabulary	3
a decision tree	3
in the context	3
open access and	3
of the work	3
of the main	3
to adapt to	3
in the data	3
in the s	3
beginning of the	3
machine learning or	3
there can be	3
data for training	3
through the library	3
has not been	3
a relational database	3
of the real	3
that feature chicago	3
an example of	3
a variety of	3
because of the	3
ethical challenges from	3
a long way	3
featured in the	3
advent of the	3
the hathitrust research	3
number of times	3
between quantitative and	3
is a simple	3
that we know	3
warmth of other	3
generator learns to	3
you may need	3
the extraction of	3
and the like	3
place names in	3
learning pilot program	3
it is still	3
about machine learning	3
any number of	3
advances in neural	3
quality of the	3
of other suns	3
of a marc	3
are interested in	3
disappointed by the	3
access to a	3
society for information	3
of the latent	3
deep learning have	3
a very large	3
a result of	3
would like to	3
have access to	3
in english and	3
question of how	3
suppose you have	3
the source data	3
of your team	3
example of a	3
in advances in	3
are good for	3
aspects of the	3
physical and virtual	3
with the advent	3
the distant reader	3
to augment knowledge	3
will have to	3
in the mag	3
an understanding of	3
that emerge from	3
to the question	3
of the more	3
commercial facial recognition	3
a marc record	3
a system that	3
as machine learning	3
if you have	3
the generator and	3
in history and	3
we found that	3
believe that the	2
the three books	2
can be reflected	2
we rely on	2
idea is to	2
in the given	2
be morally wrong	2
this raises the	2
rely on the	2
like happy marriages	2
volume of data	2
the effects of	2
of congress subject	2
as good as	2
will invite the	2
data as input	2
developing a machine	2
to work together	2
script is a	2
of institutional repositories	2
sampled to generate	2
online suicidal crises	2
like developing a	2
to use the	2
companion file to	2
books that are	2
in partnership with	2
the resulting model	2
data should be	2
personalized and automated	2
place names that	2
augmented term frequency	2
a physical body	2
of kavalier and	2
you expect to	2
the research landscape	2
to get a	2
to generate a	2
and special collections	2
developing a relationship	2
what type of	2
of the red	2
of digital humanities	2
we use machine	2
when you were	2
data should include	2
example is the	2
we did not	2
the distribution of	2
create an artificial	2
requires more than	2
automated categorization of	2
difference between modern	2
at pine mountain	2
through the process	2
of teaching and	2
submitted joint proposals	2
the scholarship of	2
pulling the lever	2
to produce realistic	2
unsupervised machine learning	2
it might be	2
different algorithms require	2
cultural heritage institutions	2
there are also	2
of the community	2
my robot overlords	2
entity type of	2
of convolutional neural	2
of your project	2
a popular example	2
to digital objects	2
creates a model	2
to create the	2
is no such	2
classroom faculty to	2
in the article	2
from references section	2
to calculate the	2
machine learning model	2
design a system	2
as we discuss	2
a labeled training	2
cultural heritage and	2
nearest whole number	2
handle multiple subjects	2
of the corpus	2
is that it	2
fulgeri et al	2
themes or topics	2
mention chicago place	2
paper presented at	2
information systems and	2
to illustrate the	2
would not be	2
of named people	2
is to help	2
weakening of cognitive	2
they deem to	2
a sequential data	2
the data upon	2
to analyze and	2
for file in	2
count tabulate the	2
also in the	2
a human in	2
and cognitive science	2
questions will help	2
learns to separate	2
has thus far	2
problems in the	2
has led to	2
part to whole	2
of archives and	2
is morally right	2
fill out this	2
please fill out	2
that rely on	2
research goals and	2
our knowledge of	2
of the first	2
hathitrust research portal	2
the end of	2
is an image	2
explicit memory can	2
the facial recognition	2
techniques in machine	2
highly personalized information	2
applications of machine	2
when this is	2
been given input	2
have designed a	2
the entire dataset	2
an attempt to	2
is like happy	2
represented in the	2
of private materials	2
library as the	2
of cognitive agency	2
the year of	2
in the language	2
as we move	2
does the library	2
this chapter outlines	2
those moral rules	2
to create their	2
it offers recommendations	2
of software or	2
a deep and	2
ml technologies and	2
what happens if	2
are trying to	2
sure the program	2
surrender of moral	2
moral deskilling in	2
that is not	2
able to solve	2
to the ai	2
practice of librarianship	2
org anthology p	2
the name of	2
your cleanup operation	2
were rounding to	2
given that our	2
in which our	2
data into a	2
the risk of	2
intelligent robotics and	2
to their students	2
a greater number	2
be capable of	2
one seems to	2
to the machine	2
core functionality of	2
was supposed to	2
experience of collaborating	2
minded to ml	2
and librarians have	2
it can help	2
depending on the	2
on to the	2
of open access	2
the needs of	2
if you were	2
table of contents	2
images of the	2
has been given	2
qualitative nature of	2
common example is	2
a way of	2
posts solicitation for	2
the scholarly canon	2
social network using	2
responses to online	2
at oklahoma state	2
and text analysis	2
can not be	2
to achieve this	2
helpfulness of peer	2
from autonomous ai	2
think it is	2
guidelines for building	2
point of view	2
libraries modules that	2
to accommodate multiple	2
challenges from autonomous	2
traditional knowledge labels	2
that it can	2
the machine to	2
applications such as	2
large amounts of	2
network using the	2
were trained on	2
is a popular	2
categorized as graph	2
for the library	2
it was a	2
from the pmss	2
both types of	2
the images and	2
you will select	2
and how it	2
flawed ai algorithms	2
see the section	2
the file with	2
you know that	2
library resources and	2
marc record is	2
the baldwin effect	2
classification and categorization	2
number of possible	2
might be welcomed	2
historians and psychologists	2
the limitations of	2
to replace the	2
for image recognition	2
ai system with	2
four subject headings	2
name is a	2
to the public	2
difference in the	2
that it would	2
as a computer	2
to keep up	2
in this context	2
up to the	2
of gans for	2
as we suggested	2
the markov chain	2
configurations until you	2
the gap between	2
be a piece	2
the first step	2
the research cycle	2
can be modeled	2
selections that feature	2
and rita cucchiara	2
can read and	2
an accuracy score	2
in creating a	2
in different types	2
number of wars	2
rather than the	2
how and why	2
chess or driving	2
is on the	2
in areas such	2
world economic forum	2
new forms of	2
principles of jus	2
beats better algorithms	2
generator and discriminator	2
for further research	2
and autonomous weapons	2
and each column	2
at the end	2
be most relevant	2
to recognize the	2
is such a	2
in relation to	2
be a lot	2
copyright and fair	2
may never have	2
good idea to	2
emerge from the	2
these ethical concerns	2
to the development	2
and preservation of	2
contributions to the	2
the libraries modules	2
not provide an	2
none of the	2
be contrasted with	2
and sojka and	2
the file path	2
welcomed by some	2
could be better	2
algorithms and machine	2
and autonomous agents	2
have the time	2
the dataset is	2
ltr issue viewissue	2
a more accessible	2
insert reference here	2
we are still	2
never have the	2
hardware and software	2
the history of	2
integrated into a	2
in the early	2
ready to go	2
just about any	2
to change the	2
dataset for creation	2
in the scholarly	2
to accept the	2
should not be	2
adding more fields	2
activities or narrowly	2
the data is	2
of human activities	2
tools programming onebook	2
i have to	2
likely to mediate	2
that could be	2
in this example	2
learning describes algorithms	2
in library and	2
from multiple sources	2
the authors column	2
the digital assistant	2
in the learning	2
nothing to do	2
of your process	2
by per y	2
by the difference	2
in ieee cvf	2
structure of a	2
learning applications could	2
that does not	2
images of people	2
they are also	2
and present an	2
important to note	2
a machine does	2
of technical literature	2
of a larger	2
kavalier and clay	2
a work in	2
reference service to	2
us to continue	2
right thing to	2
fake and real	2
analysis of the	2
working with immutable	2
denoted as a	2
the idea that	2
file for file	2
data set is	2
the united states	2
is also the	2
the three most	2
the data are	2
can play a	2
of the original	2
to do this	2
to believe that	2
a site of	2
contents of the	2
contrasted with a	2
to match the	2
interests in privacy	2
i know about	2
to stop worrying	2
of using computers	2
would be to	2
of computers is	2
a relatively recent	2
in turn could	2
for a more	2
as playing chess	2
on your p	2
thank you for	2
voice user interface	2
first of all	2
the domain expert	2
control of their	2
the point is	2
and meng jiang	2
led a team	2
we had coffee	2
if you had	2
data might be	2
archivists and librarians	2
be allowed to	2
work as preliminary	2
software or a	2
com ageitgey face	2
and ai in	2
and preserve the	2
to provide the	2
focus on the	2
in recent years	2
many ai researchers	2
on an english	2
the matrix and	2
text analysis tools	2
but what if	2
the sentiment associated	2
machines can be	2
the idea is	2
a common example	2
what features were	2
existing beliefs and	2
the results were	2
in a text	2
relevant to your	2
in one place	2
of peer support	2
as military robots	2
and whether or	2
for historians to	2
proquest ebook central	2
were going to	2
the aws application	2
process used to	2
the nearest tenth	2
place in the	2
name and a	2
cleanup and transformation	2
or not you	2
to a new	2
means an exhaustive	2
the early stage	2
the nearest whole	2
close to a	2
of ai systems	2
one or more	2
for the institution	2
weapons of math	2
be involved in	2
can help you	2
model based on	2
the phenomenon that	2
is common in	2
is what makes	2
were deemed revolutionary	2
or a machine	2
cognitive agency and	2
with the ability	2
a better understanding	2
we are not	2
read from beginning	2
performance can be	2
use the library	2
the automated categorization	2
learning solution to	2
well as to	2
how i learned	2
characteristics of a	2
a best practice	2
have also submitted	2
a more nuanced	2
the extent that	2
for consumer research	2
ai applications for	2
also important to	2
in the area	2
the identity of	2
based upon a	2
and testing sets	2
how much of	2
and yoshua bengio	2
out this reference	2
to report on	2
but this is	2
as the generator	2
is not that	2
then be sampled	2
values of the	2
a form of	2
use of digital	2
the practice of	2
to design a	2
much as possible	2
on the left	2
key information or	2
about how we	2
on descriptive metadata	2
help you to	2
work with other	2
opportunities to augment	2
ability to save	2
they relate to	2
bogged down with	2
on our work	2
we compared the	2
will do the	2
would be able	2
you can find	2
we have been	2
cited on your	2
modules that will	2
data may not	2
function in the	2
of possible outcomes	2
a review of	2
university yearbook collections	2
as a tool	2
an algorithm will	2
the types of	2
are not commonly	2
suppose there is	2
the tension between	2
order markov chain	2
outputs of different	2
rise of the	2
data literacy to	2
that we have	2
take advantage of	2
letters in english	2
were working with	2
in the age	2
a response to	2
deem to be	2
archivists and curators	2
reaching out to	2
cai et al	2
you will need	2
solution to facilitate	2
for more on	2
a piece of	2
tried to use	2
the skills and	2
evaluate the quality	2
will have a	2
to human understanding	2
the right hammer	2
the real problem	2
and a list	2
the connections between	2
human activities or	2
van den dool	2
a deep learning	2
that we did	2
i do not	2
a bit more	2
of her as	2
to make ai	2
from other literature	2
you can often	2
ageitgey face recognition	2
patients with both	2
for this reason	2
date and timestamp	2
did most of	2
of the humanities	2
also be a	2
such as python	2
you plan to	2
biases and the	2
used to train	2
without having to	2
your use case	2
this can include	2
pedagogy in library	2
in the loop	2
you for addressing	2
machine learning algorithm	2
supplement bibliographic description	2
of early chinese	2
goal was to	2
a gun with	2
extraction of location	2
per x was	2
of capturing the	2
is the use	2
congress posts solicitation	2
on a dvd	2
indexes as a	2
if they are	2
social implications of	2
for the purpose	2
machine learning models	2
i write this	2
and the ml	2
that is as	2
to create broader	2
light of these	2
of public trust	2
of notre dame	2
to be read	2
learning have brought	2
list of sentences	2
or four subject	2
machine learning application	2
to the matrix	2
other cultural institutions	2
trained on a	2
more about the	2
your ml process	2
the members of	2
org tools programming	2
plan to use	2
of times a	2
the success of	2
for addressing our	2
of the works	2
digital assistants are	2
of hidden biases	2
trained on an	2
information technology libraries	2
and why it	2
exactly the same	2
we came to	2
msc values to	2
no such thing	2
the topic modeling	2
use of computers	2
discussion of the	2
and dissemination of	2
some of these	2
not able to	2
through machine learning	2
focused on developing	2
it possible for	2
you train a	2
world of machine	2
if you know	2
appears to be	2
to ml technologies	2
to produce better	2
broadly in all	2
willing to create	2
the purposes of	2
to generate an	2
to identify and	2
instance as our	2
the github repository	2
hundred and forty	2
in a classification	2
not have developed	2
general purpose intelligence	2
regions of the	2
chicago of fiction	2
head of ai	2
of labeled data	2
the place names	2
instead of trying	2
and did not	2
our work in	2
peer responses to	2
and weaknesses of	2
version of a	2
a moral theory	2
choose an algorithm	2
application of results	2
argues that the	2
of the covid	2
autonomous weapons systems	2
in the stanford	2
following python script	2
research is like	2
for privacy and	2
to make sure	2
named entity recognition	2
such as storing	2
towards data science	2
us department of	2
labeled training dataset	2
version of the	2
algorithm will not	2
finding the right	2
images on the	2
edited by patrick	2
also known as	2
but it is	2
way that it	2
of speech recognition	2
as a whole	2
of a non	2
well as their	2
ml research to	2
not violate copyright	2
scholars to add	2
acm sigkdd international	2
wen et al	2
of decision trees	2
the amazing adventures	2
of ai research	2
going to revolutionize	2
bibliographic descriptions of	2
supports iteration and	2
want to test	2
international committee of	2
scholarship of teaching	2
to be most	2
with a larger	2
the probabilities of	2
enable researchers to	2
the near future	2
and the data	2
or data science	2
your data as	2
a version of	2
understand the concept	2
research that we	2
also be used	2
in the sense	2
they would be	2
in the public	2
to achieve a	2
in its performance	2
success or failure	2
and petr sojka	2
datasets that are	2
a specific activity	2
provide machine learning	2
to online suicidal	2
the field will	2
to get to	2
find that the	2
the articles categorized	2
be incorporated into	2
at the intersection	2
artificial intelligence is	2
not as important	2
testing and training	2
can be shared	2
be a good	2
the output of	2
note that by	2
libraries can play	2
seattle university law	2
between learning and	2
the scale of	2
technologies and the	2
text of articles	2
a linear model	2
examples and the	2
scholarly record and	2
the cost of	2
those digital assistants	2
the qualitative nature	2
with a few	2
moral intuition in	2
is an area	2
to predict an	2
ethics does not	2
on new variations	2
does not provide	2
to rely on	2
american library association	2
the middle ages	2
what a machine	2
generate new images	2
on a known	2
is a topic	2
in some way	2
at the work	2
we will invite	2
the three steps	2
read and write	2
the advantages of	2
a deeper understanding	2
of the existing	2
on sets of	2
in ml x	2
arrived at an	2
of local communities	2
individual research work	2
being able to	2
the section on	2
solve this problem	2
university of pretoria	2
keeping track of	2
no way to	2
be shared with	2
datasets that can	2
php ltr issue	2
is artificial intelligence	2
agency and autonomy	2
key information from	2
is a sequential	2
about the conclusion	2
is given context	2
learn about the	2
new knowledge through	2
save time and	2
this is also	2
you what the	2
and love my	2
images of skin	2
across the corpus	2
the model is	2
which they are	2
be read from	2
out that the	2
assigned to the	2
social discord and	2
quantities of data	2
are only as	2
gun with a	2
to define what	2
with one another	2
in the last	2
metaphysics research lab	2
and the generated	2
new approach to	2
of many different	2
both managing and	2
your data to	2
list of directories	2
libraries have not	2
middle and high	2
adventures of kavalier	2
role in both	2
open source software	2
to use a	2
increases power of	2
the audiovisual metadata	2
aim to make	2
such as playing	2
versions of the	2
popular example of	2
file to each	2
an emphasis on	2
of collections that	2
for facial recognition	2
sigkdd international conference	2
or narrowly in	2
are examples of	2
dialogue with our	2
strengths of happy	2
only way to	2
classification and regression	2
that they were	2
screening mammography with	2
all areas of	2
you will use	2
of library resources	2
this kind of	2
are also being	2
of the time	2
how well the	2
the resulting topics	2
our use of	2
to explain what	2
to spend time	2
state of the	2
making machine learning	2
to the researcher	2
depending on how	2
to enhance library	2
weights and biases	2
data structures have	2
the functionality of	2
compare the results	2
require the libraries	2
are relevant to	2
loop through each	2
and it can	2
ethical challenges that	2
likely to suffer	2
studies in history	2
if the model	2
can be fooled	2
is also possible	2
playing chess or	2
a mathematical model	2
represents a document	2
of the grants	2
state university yearbook	2
be an important	2
is only a	2
solve the problem	2
go beyond the	2
to make a	2
possible use cases	2
the ml expert	2
coded into the	2
no means an	2
not having the	2
and development phase	2
to the number	2
evaluate the resulting	2
make it clear	2
is willing to	2
and datasets that	2
would you expect	2
to enrich the	2
look at the	2
in the images	2
is by no	2
describes algorithms that	2
local histories of	2
the possibility of	2
the mathematics articles	2
powered digital assistants	2
enough to produce	2
handle datasets with	2
not only by	2
will only increase	2
the images we	2
that makes the	2
useful can be	2
other cultural heritage	2
began to look	2
method to the	2
could only be	2
is also important	2
the domain of	2
machine learning deep	2
sketches from google	2
learning instruction and	2
the problem is	2
model to classify	2
of input data	2
model has been	2
may argue that	2
to apply these	2
sometimes you have	2
we would like	2
opaque to human	2
that will do	2
noise in the	2
is known about	2
a large number	2
in a narrow	2
data as immutable	2
convolutional neural nets	2
may result in	2
the library profession	2
how well it	2
if you train	2
two markov chains	2
throughout the course	2
whether it is	2
to employ machine	2
with a physical	2
york public library	2
data from the	2
would be a	2
in the machine	2
act in a	2
library spaces to	2
the outputs of	2
circulation of a	2
lack of a	2
creating a culture	2
to evaluate the	2
managing and creating	2
and the classifier	2
number of documents	2
the type of	2
memorandum of understanding	2
out to be	2
such as network	2
of happy marriages	2
marriage family review	2
for the categorization	2
in a variety	2
to have more	2
the sense that	2
after the first	2
lifetime adaptations are	2
it is hard	2
back into the	2
has become a	2
it would help	2
into a photo	2
and raj reddy	2
this is by	2
and more capable	2
assistants are likely	2
and a directory	2
file with open	2
computer science and	2
depending on your	2
basic reference service	2
available a curated	2
for object detection	2
ma et al	2
to new data	2
sentiment associated with	2
good example of	2
size of the	2
stop worrying and	2
using generative adversarial	2
and maybe a	2
a data structure	2
and make it	2
is not clear	2
development of new	2
machine learning would	2
reflected in cross	2
of your cleanup	2
with a very	2
analysis tools and	2
to the project	2
committee of the	2
trying to replace	2
see if the	2
three oboc selections	2
support of curation	2
to access and	2
of a pipeline	2
knowledge discovery and	2
the oklahoma state	2
and remote sensing	2
book indexes as	2
age of artificial	2
using the contentdm	2
learning algorithm that	2
the divergence of	2
not just in	2
be of the	2
an ethical decision	2
in the later	2
sentences extracted from	2
divided into three	2
parts of your	2
for archives and	2
of this paper	2
to be more	2
retrieval activities in	2
language of mathematics	2
really understand the	2
in that light	2
we will discuss	2
on entity retrieval	2
at berea college	2
are satisfied with	2
will not be	2
possible machine learning	2
be surprised if	2
play a dynamic	2
the focus of	2
as the model	2
to gain a	2
used machine learning	2
data beats better	2
how can we	2
libraries do not	2
the most part	2
at some point	2
intelligence and the	2
the future as	2
workshop on entity	2
and the human	2
association for consumer	2
from the library	2
it was trained	2
as plain text	2
the result is	2
idea of sharing	2
as preliminary results	2
taken into account	2
a dynamic role	2
project aims to	2
public trust and	2
satisfied with communication	2
indexing and retrieval	2
challenges of cross	2
of this essay	2
for the use	2
digital and computational	2
create broader impact	2
conference on knowledge	2
and unsupervised learning	2
is a serious	2
in medical imaging	2
and bibliographic descriptions	2
words and formulae	2
hathitrust digital library	2
in both managing	2
machine learning for	2
is trying to	2
learning deep learning	2
libraries and the	2
and creating data	2
that we won	2
the rise of	2
research and development	2
ml x collaboration	2
can help to	2
association for information	2
became much more	2
in the beginning	2
of the discriminator	2
the problem with	2
success of that	2
to ai systems	2
to automate the	2
was a breakthrough	2
which would allow	2
j v n	2
association for computing	2
score greater than	2
does not have	2
economic and social	2
you know the	2
of digital collections	2
modern technology and	2
such a case	2
what would you	2
i think readers	2
enhance library collections	2
to be used	2
such as data	2
changes in the	2
you will likely	2
employ machine learning	2
of the dataset	2
digital collections in	2
the nd annual	2
will want to	2
we have found	2
is much improved	2
and learn from	2
will be the	2
our reading habits	2
with functional morality	2
academic increases power	2
edited by edward	2
if your data	2
pmss archival materials	2
applications of gans	2
digital scholarship projects	2
you prepare your	2
text files and	2
and creative practices	2
is not to	2
of math destruction	2
they wanted to	2
allows us to	2
the process that	2
for a long	2
my clean data	2
this as a	2
a subset of	2
social impact of	2
your machine learning	2
two student interns	2
of data that	2
have begun to	2
to be the	2
directory another directory	2
impact of scholarship	2
the poor health	2
the chicago public	2
labeled and unlabeled	2
that we can	2
you are using	2
this means that	2
advantages of a	2
intelligence in the	2
sys sanity check	2
whether a place	2
able to know	2
for use in	2
as a guide	2
department of justice	2
the mechanisms of	2
one can imagine	2
amazing adventures of	2
creating data and	2
ethical and social	2
subset of the	2
bigotry hampering civic	2
in the trolley	2
input to a	2
both supervised and	2
is as intelligent	2
program that will	2
while this can	2
in order for	2
failure of the	2
learning and text	2
of congress posts	2
deeper understanding of	2
feature as long	2
from a large	2
in each run	2
our collaboration has	2
tolosana et al	2
even though we	2
accuracy of digital	2
beginning of this	2
state university archives	2
with immutable data	2
is not only	2
so that it	2
go back to	2
identified all of	2
the original data	2
is mentioned in	2
a simple classification	2
ethical implications of	2
ai and its	2
the first month	2
ml techniques can	2
locality of the	2
particular areas of	2
minimize institutional memory	2
using such a	2
that could not	2
suggest library resources	2
with a focus	2
cognitive agency is	2
labels into training	2
for creation of	2
how libraries can	2
the scholarly record	2
a succeeding letter	2
they are not	2
history and philosophy	2
is improved to	2
it is then	2
is machine learning	2
by no means	2
to the other	2
semantic search by	2
image of a	2
look forward to	2
and italian dictionary	2
applications as a	2
and the student	2
of the strengths	2
how they relate	2
trust and civic	2
gans have been	2
to build many	2
create a new	2
different data structures	2
training dataset for	2
another example of	2
to work on	2
search by adding	2
designing an ai	2
full text and	2
given input if	2
to augment our	2
that mention chicago	2
data and datasets	2
chains trained on	2
has been used	2
but also in	2
within the context	2
the book thief	2
hollywood intelligent machines	2
the characteristics of	2
of data structures	2
scholarship and research	2
be an opportunity	2
in the next	2
twenty years ago	2
a library may	2
used as a	2
the course of	2
the five people	2
i think it	2
encyclopedia of philosophy	2
shared with the	2
categorization of highly	2
discovery and data	2
neural networks are	2
to have access	2
university of oklahoma	2
new york public	2
seen in the	2
systems and tools	2
you were rounding	2
to deal with	2
cases where the	2
of the file	2
congress subject headings	2
human labor to	2
number of sentences	2
is to create	2
in terms of	2
learn more about	2
the core of	2
are disappointed by	2
and create knowledge	2
like a nail	2
creation of new	2
articles categorized as	2
new learning paradigm	2
the rate of	2
per y per	2
have not yet	2
expect to have	2
and computational humanities	2
to note that	2
face recognition models	2
mammography with and	2
in many different	2
from the archives	2
a curated list	2
members of local	2
activities in the	2
their impact on	2
a database or	2
that the results	2
learning and training	2
banerjee et al	2
are many possible	2
was written by	2
file name and	2
or failure of	2
faster and more	2
within the archives	2
ml expert is	2
at the early	2
learns how to	2
help you make	2
supervised machine learning	2
theses and dissertations	2
cvf conference on	2
emerged from a	2
data are good	2
examples that are	2
building an alexa	2
review of the	2
marc yyyymmdd hhmmss	2
context of the	2
can be very	2
two examples of	2
only by the	2
could be done	2
at scale to	2
the area of	2
it was also	2
of research papers	2
of oklahoma libraries	2
are to be	2
able to do	2
the most common	2
metadata creation in	2
your file path	2
encoded in the	2
press releases are	2
place name data	2
amount of information	2
chicago location is	2
an english and	2
whether they were	2
memory can be	2
of gans that	2
of a succeeding	2
of these methods	2
at the outset	2
and the reality	2
on knowledge discovery	2
in unsupervised learning	2
learning was supposed	2
of the social	2
of a system	2
digital humanities could	2
the generated data	2
are far from	2
compare the outputs	2
accept the apple	2
and output the	2
the authors of	2
industries that preceded	2
and peer posts	2
the entire ml	2
data structure designed	2
of the images	2
a good idea	2
for computing machinery	2
will be using	2
of my data	2
of a labeled	2
do not discuss	2
understand what is	2
and george a	2
they are able	2
the same pattern	2
of the ethical	2
seem to be	2
sequential data structure	2
the only way	2
intelligence and policing	2
all aspects of	2
is the only	2
groups at the	2
a method of	2
of moral agency	2
and its people	2
and you want	2
such as machine	2
expert is willing	2
vast amounts of	2
the weather is	2
the implementation of	2
good for training	2
accuracy of the	2
that we will	2
to the extent	2
a tool that	2
of this chapter	2
and philosophy of	2
in robot ethics	2
the archive and	2
will allow us	2
order to achieve	2
since machine learning	2
in the field	2
an artificial system	2
can be contrasted	2
deep fakes are	2
in the document	2
which parts are	2
machine learning instruction	2
deep learning is	2
collaborators and other	2
helpful peer responses	2
subject headings to	2
data and labels	2
learning in a	2
rely on a	2
make available a	2
as a place	2
looking at the	2
a process that	2
phenomenon of learning	2
the former director	2
of jus in	2
for i in	2
digital assistants become	2
you are working	2
text and bibliographic	2
might not be	2
problem is that	2
on your dataset	2
power of semantic	2
such as neural	2
and the quality	2
time alone and	2
nd annual meeting	2
new training set	2
and the discriminator	2
during the initial	2
service to their	2
have a set	2
our research groups	2
known about the	2
lehman et al	2
and fair use	2
in the near	2
work in progress	2
us to focus	2
when classifying the	2
large corpora of	2
the documents into	2
greater computing capabilities	2
iteration and experimentation	2
their existing beliefs	2
that uses machine	2
ieee international conference	2
of an action	2
is the year	2
facilitate metadata creation	2
applied at scale	2
also want to	2
policies and regulations	2
ai in libraries	2
with generative adversarial	2
ieee conference on	2
preservation of knowledge	2
is an integral	2
world use cases	2
of directories containing	2
piece of software	2
outside of the	2
text of the	2
the reality of	2
dynamic role in	2
also submitted joint	2
realistic images of	2
the training set	2
x was taught	2
the faces of	2
of the automated	2
when we had	2
is a value	2
about the world	2
level information about	2
it be a	2
and testing models	2
text mining and	2
the library provides	2
new labeled data	2
areas of human	2
to facilitate metadata	2
computing infrastructure and	2
implications of robotics	2
in complex ways	2
of having to	2
descriptions of all	2
have brought significant	2
ai will be	2
in archives and	2
step until satisfied	2
was not the	2
machine learning could	2
in spite of	2
at the heart	2
digital collections by	2
eric lease morgan	2
of classifying mathematics	2
the th acm	2
not always be	2
the entire sentence	2
by adding more	2
the category of	2
as intelligent as	2
or a string	2
for the purposes	2
systems that are	2
each data output	2
used in a	2
the associations between	2
more fields of	2
a type of	2
physical and digital	2
that can serve	2
with the members	2
the two main	2
the part of	2
the classification of	2
the stanford encyclopedia	2
each column is	2
in the scope	2
have a large	2
to save time	2
deep learning can	2
and so on	2
an alexa application	2
you can use	2
in this sense	2
facilitate open access	2
the association between	2
they do not	2
deep reinforcement learning	2
for training and	2
invite the public	2
creation in support	2
very large datasets	2
lot is known	2
and look for	2
we knew that	2
an opportunity for	2
so it is	2
long way from	2
the library website	2
until you get	2
by deep learning	2
know about this	2
is facilitated by	2
of scholarship and	2
opacity in decision	2
preservation and access	2
to strike a	2
in the creation	2
greater number of	2
from which the	2
which in turn	2
highly technical content	2
the principles of	2
please insert reference	2
to do so	2
her as a	2
some sort of	2
of learning in	2
whether it be	2
same set of	2
not likely to	2
can handle multiple	2
the companion file	2
as a wheel	2
of generative learning	2
using computers to	2
they can be	2
beginning to end	2
next step in	2
is a method	2
with the data	2
art and design	2
can see that	2
would only be	2
not have the	2
human in its	2
kentucky public schools	2
questions about the	2
that you need	2
create new knowledge	2
education and research	2
with historians and	2
for the most	2
the case for	2
hosted a summit	2
of digital screening	2
should one handle	2
all forms of	2
path for the	2
still a long	2
generated words are	2
place names were	2
a specific algorithm	2
of the algorithm	2
oboc books that	2
with the library	2
using deep learning	2
a chicago location	2
may represent a	2
turned out that	2
on what i	2
i in range	2
ml algorithms are	2
facial recognition software	2
used to build	2
hope is that	2
the input can	2
that go beyond	2
a matrix of	2
be easy to	2
a right thing	2
answer the question	2
the ml process	2
results of the	2
to support the	2
stanford encyclopedia of	2
each and every	2
of these processes	2
algorithms require different	2
this could be	2
of skin lesions	2
either broadly in	2
together based on	2
curated list of	2
decisions and actions	2
university of illinois	2
and social implications	2
the party that	2
bring people together	2
labeled place name	2
the first of	2
hampering civic discourse	2
that we were	2
be sampled to	2
is in a	2
the matrix on	2
ensures that you	2
support vector machines	2
a method to	2
standard supervised learning	2
we have begun	2
a new machine	2
as black boxes	2
it must be	2
uncover helpful peer	2
its moral concerns	2