U.S. patent application number 13/912368 was filed with the patent office on 2014-12-11 for system and method of improving communication in a speech communication system.
The applicant listed for this patent is Jurgen Totzke. Invention is credited to Jurgen Totzke.
Application Number | 20140365213 13/912368 |
Document ID | / |
Family ID | 52006208 |
Filed Date | 2014-12-11 |
United States Patent
Application |
20140365213 |
Kind Code |
A1 |
Totzke; Jurgen |
December 11, 2014 |
System and Method of Improving Communication in a Speech
Communication System
Abstract
A speech communication system and a method of improving
communication in such a speech communication system between at
least a first user and a second user may be configured so the
system (a) transcribes a recorded portion of a speech communication
between the at least first and second user to form a transcribed
portion, (b) selects and marks at least one of the words of the
transcribed portion which is considered to be a keyword of the
speech communication, (c) performs a search for each keyword and
produces at least one definition for each keyword, (d) calculates a
trustworthiness factor for each keyword, each trustworthiness
factor indicating a calculated validity of the respective
definition(s), and (e) displays the transcribed portion as well as
each of the keywords together with the respective definition and
the trustworthiness factor thereof to at least one of the first
user and the second user.
Inventors: |
Totzke; Jurgen; (Poing,
DE) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Totzke; Jurgen |
Poing |
|
DE |
|
|
Family ID: |
52006208 |
Appl. No.: |
13/912368 |
Filed: |
June 7, 2013 |
Current U.S.
Class: |
704/235 |
Current CPC
Class: |
G10L 2015/088 20130101;
G06F 40/232 20200101; G06F 40/284 20200101; G10L 15/06 20130101;
G06F 40/30 20200101; G10L 15/26 20130101; G10L 15/1822 20130101;
G10L 15/22 20130101; G06F 3/04812 20130101; G10L 15/08 20130101;
G06F 40/242 20200101; G06Q 10/10 20130101; G06F 3/0482 20130101;
G10L 21/02 20130101; G10L 15/1815 20130101; G10L 21/10
20130101 |
Class at
Publication: |
704/235 |
International
Class: |
G10L 15/26 20060101
G10L015/26 |
Claims
1. A method of improving communication in a speech communication
system between at least a first user and a second user, the method
comprising: a) the system transcribing a recorded portion of a
speech communication between the at least first and second user to
form a transcribed portion, b) the system selecting and marking at
least one of the words of the transcribed portion which is
considered to be a keyword of the speech communication, c) the
system performing a search for each keyword and produces at least
one definition for each keyword, d) the system calculating a
trustworthiness factor for each keyword, each trustworthiness
factor indicating a calculated validity of the respective
definition(s), and e) the system displaying the transcribed portion
as well as each of the keywords together with the respective
definition and the trustworthiness factor thereof to at least one
of the first user and the second user.
2. The method of claim 1, wherein step a) comprises using a
real-time recording of the speech communication.
3. The method of claim 1, wherein step a) comprises a spell-check
of the transcribed portion using at least one regular and/or at
least one domain glossary.
4. The method of claim 1, wherein step b) comprises an automatic
selecting of the keywords by the system and manually selecting of
the keywords by a user and/or by a community of users using the
system.
5. The method of claim 1, wherein step c) comprises a search in a
structured and/or in an unstructured data repository.
6. The method of claim 1, wherein step d) comprises a semantic
search using ontologies.
7. The method of claim 1, wherein step d) comprises a semantic
search using taxonomies in order to generate a taxonomy correction
factor for modifying the trustworthiness factor.
8. The method of claim 1, wherein step d) comprises carrying out a
similarity checking step which takes into account the similarity of
various definitions of a respective keyword in order to generate a
similarity correction factor for modifying the trustworthiness
factor.
9. The method of claim 1, wherein step d) comprises carrying out a
step of sentiment detection which takes into account the sentiment
with respect to at least one definition of at least one keyword in
order to generate a sentiment correction factor for modifying the
trustworthiness factor.
10. The method of claim 1, wherein step a) comprises translating
the transcribed portion into a pre-defined language.
11. The method of claim 1, wherein wherein the speech communication
results in a communication thread and wherein a glossary of the
communication thread is created and regularly updated
12. A non-transitory computer-readable medium comprising a computer
program that defines a method that is performed by a communication
system when the system runs the program, the method comprising: a)
the system transcribing a recorded portion of a speech
communication between the at least first and second user to form a
transcribed portion, b) the system selecting and marking at least
one of the words of the transcribed portion which is considered to
be a keyword of the speech communication, c) the system performing
a search for each keyword and produces at least one definition for
each keyword, d) the system calculating a trustworthiness factor
for each keyword, each trustworthiness factor indicating a
calculated validity of the respective definition(s), and e) the
system displaying the transcribed portion as well as each of the
keywords together with the respective definition and the
trustworthiness factor thereof to at least one of the first user
and the second user.
13. A speech communication system, comprising: a transcription
unit, the transcription unit transcribing a recorded portion of a
speech communication between the at least first and second user to
form a transcribed portion, a selecting unit, the selecting unit
selecting and marking at least one of the words of the transcribed
portion which is considered to be a keyword of the speech
communication, a search unit, the search unit performing at least
one search for each keyword and producing at least one definition
for each keyword, a trustworthiness unit, the trustworthiness unit
calculating a trustworthiness factor for each keyword, each
trustworthiness factor indicating a calculated validity of the
respective definition(s), and a display unit, the display unit
displaying the transcribed portion as well as each of the keywords
together with the respective definition and the trustworthiness
factor thereof to at least one of the first user and the second
user.
14. The speech communication system of claim 13, further comprising
a recording unit, the recording unit recording real-time speech
communication.
15. The speech communication system of claim 13, further comprising
a spell-check unit, the spell-check unit spell checking the
transcribed portion using at least of a regular domain glossary and
a domain glossary.
16. The speech communication system of claim 13, wherein the
selection unit automatically selects at least one of the words
based on one of the keywords and the system further comprising an
input device such that a manual selecting of the keywords by at
least one user is inputtable to the system.
17. The speech communication system of claim 13, further comprising
a search unit, the search unit searching in a structured and/or in
an unstructured data repository.
18. The speech communication system of claim 13, further comprising
a semantic search unit, the semantic search unit performing a
semantic search using ontologies.
19. The speech communication system of claim 13, further comprising
a semantic search unit, the semantic search unit performing a
semantic search using taxonomies in order to generate a taxonomy
correction factor for modifying the trustworthiness factor.
20. The speech communication system of claim 13, further comprising
a similarity check unit, the similarity check unit carrying out a
similarity checking step which takes into account the similarity of
various definitions of a respective keyword in order to generate a
similarity correction factor for modifying the trustworthiness
factor.
21. The speech communication system of claim 13, further comprising
a sentiment check unit, the sentiment check unit carrying out a
step of sentiment detection which takes into account the sentiment
with respect to at least one definition of at least one keyword in
order to generate a sentiment correction factor for modifying the
trustworthiness factor.
22. The speech communication system of claim 13, further comprising
a translation unit, the translation unit translating the
transcribed portion into a pre-defined language.
23. The speech communication system of claim 13, further comprising
a data storage unit, the data storage unit storing a communication
thread resulting from the speech communication results and storing
a glossary created from the communication thread.
Description
FIELD OF THE INVENTION
[0001] The present invention relates to a method of improving
communication in a speech communication system between at least one
first user and one second user. Furthermore, the present invention
relates to a speech communication system for implementing such a
method.
BACKGROUND OF THE INVENTION
[0002] Knowledge workers are expected to provide their expertise to
many different projects within their company. In many cases this
results in the fact that they are involved in various teams, which
may also be virtual teams. They typically have to attend one
teleconference or collaboration session after the other without
really having enough time to prepare themselves completely for the
current project context.
[0003] In other instances, the knowledge worker just relaxes
between the previous telephone conference and the next one. Assume
she or he is going to discuss in a telephone conference the new
business opportunity from an emerging domain that she/he is not
familiar with at all. Due to a lack of time she/he is not
well-prepared. Nevertheless, it is important for her/him to show
competences during the telephone conference in order to anticipate
their participation in the project.
[0004] I have determined that it appears that knowledge workers
have not yet enough support by the speech communication system they
are using for facing such problems.
SUMMARY OF THE INVENTION
[0005] Consequently, it is an object of the present invention to
provide a method and a speech communication system which provides a
better support for knowledge workers in situations like the ones
discussed above, whereby the communication process may be improved
and the efficiency of the knowledge worker may be increased. Thus,
the knowledge worker should receive a maximum of support from the
system with a minimum of personal cognitive effort. The object is
in other words to provide all the relevant information and context
to the persons involved in such a communication process.
[0006] At present, during an on-going speech communication,
knowledge workers may mainly apply an unstructured procedure for
retrieving additional information, such as searching on the
Internet or in certain information spaces or in e-mail archives.
These data sources may be called data repositories in a
generalizing manner. Consequently, the attention of the
participants in such a speech communication may get heavily
distracted from the main issue of this speech communication, namely
the topic to be discussed. It is to be emphasized that the term
"speech communication" is to be intended that it refers to any
communication process where speech is a part of this process.
Examples of such a speech communication are audio communications
such as a telephone call or teleconference, or a video
communication such as a video conference.
[0007] This problem is solved with a method of improving
communication in a speech communication system according to claim
1, comprising the steps that the system:
a) transcribes a recorded portion of a speech communication between
the at least first and second user to form a transcribed portion,
b) selects and marks at least one of the words of the transcribed
portion which is considered to be a keyword of the speech
communication, c) performs a search for each keyword and produces
at least one definition for each keyword, d) calculates a
trustworthiness factor for each keyword, each trustworthiness
factor indicating a calculated validity of the respective
definition(s), and e) displays the transcribed portion as well as
each of the keywords together with the respective definition and
the trustworthiness factor thereof to at least one of the first
user and the second user.
[0008] The term "portion of a speech communication" is to be
understood such that also a complete speech communication may be
considered and not only a part of it, as the use case may be.
[0009] According to a further aspect of the present invention, this
problem can also be solved by a speech communication system
according to claim 13, comprising the following functional
entities: [0010] a transcription unit for transcribing a recorded
portion of a speech communication be-tween the at least first and
second user to form a transcribed portion, [0011] a marking unit
for selecting and marking at least one of the words of the
transcribed portion which is considered to be a keyword of the
speech communication, [0012] a search unit for performing at least
one search for each keyword and producing at least one definition
for each keyword, [0013] a trustworthiness unit for calculating a
trustworthiness factor for each keyword, each trustworthiness
factor indicating a calculated validity of the respective
definition(s), and [0014] a display unit for displaying the
transcribed portion as well as each of the keywords together with
the respective definition and the trustworthiness factor thereof to
at least one of the first user and the second user.
[0015] Respective advantageous embodiments of the invention are
subject-matter of the dependent claims.
[0016] Definitions of terms used with respect to this
invention:
[0017] Similarity is defined as the semantic similarity, whereby a
set of terms within term lists are evaluated on the likeness of
their meaning/semantic content.
[0018] Ontology as defined for computer and information science
formally represents knowledge within a domain. Ontologies provide a
shared vocabulary, which can be used to model a domain with the
type of objects and their properties and relations. Ontology
organizes information as a form of knowledge representation about a
domain. The Web Ontology Language (OWL) as defined by W3C is a
family of knowledge representation languages for authoring
ontologies.
[0019] Taxonomy applied to information science is a hierarchical
structure of classified objects. A taxonomy can be regarded as a
special, simplified ontology for which the relations of objects are
hierarchical.
[0020] Sentiment Detection (also known as Sentiment Analysis or
Opinion Mining) refers to the application of natural language
processing, computational linguistics, and text analytics to
identify and extract subjective information in source
materials.
[0021] Embodiments of the invention may cover, among other things,
the following aspects: While a user is in a telephone conference
the user may activate the disclosed method and system. A window may
pop up showing the real-time transcription of the spoken words.
Nouns and terms are automatically detected and marked. With a
background application structured and unstructured information from
internal and external data sources may be searched and
consolidated. If augmenting information can be provided out of
these search results the text gets highlighted. On mouse-over the
augmented information is displayed. Alternatively, high-lighting
transcription text by the user activates this function
manually.
[0022] Based on the search results, the grade of trustworthiness
(i.e. veracity or validity) of the provided, consolidated
information is estimated (i.e. calculated) and displayed applying
technologies like semantic similarity (or semantic relatedness),
whereby the likeness of terms in respect to their meaning/semantic
content is determined, and sentiment is detected.
[0023] During a subsequent teleconference and collaboration session
out of a series, augmented transcriptions from the previous
sessions may be displayed in backward chronological order. While
scrolling down, the user can rapidly remember the project context,
her/his intended contributions, and facilitate consistent
positioning to the ongoing discussion. If certain terms and
references are from a domain with which the user is not familiar
with, the user typically doesn't want to consume the time of the
other domain experts. Thanks to this invention, highlighted terms
provide you with definitions and context, e.g. on mouse-over.
High-lighting may apply automatically, e.g. special terms which are
typically outside the regular dictionary, or selected by other
members of the collaboration team. Most frequent mentioned terms
from the previous discussion or session series are presented in an
automatically generated tag-cloud providing a flashlight on the
problem space discussed.
[0024] The invention applies structured and unstructured search,
and semantic similarity, to online transcription of a
conference/collaboration session complemented with a
trustworthiness indication to create a contextual communication
experience. Furthermore, the disclosed embodiments allow for
playback of augmented transcriptions from concluded sessions or
ongoing series of conferencing/collaboration sessions.
[0025] Embodiments of the speech communication system and the
corresponding method of this invention may comprise the following
features. Communication systems comprise audio/video conference
units for media mixing, recording, and streaming of media. The
transcripting function transcribes the audio component of the media
into a textual representation. Typically, the transcript contains
errors that can be auto-corrected by applying a spell-checker
function using a regular dictionary for the language in use.
Remaining errors that could not be resolved will be matched against
a domain specifis dictionary/glossary. If the error can be
resolved, the related information is retrieved and linked to the
transcript. Otherwise, the term will be marked and high-lighted by
the word spotting functional entity. Spotted words are then applied
to a search at information spaces listed in a trusted information
space directory. Items in this directory are accompanied with a
trustworthiness factor related to the information space searched
and the type of search applicable. The directory includes
references to pre-defined information spaces that are applicable
for structured or unstructured search, e.g. well-known information
source like Wikipedia, or applying semantic search typically for
information available in intranet or a data warehouse, or for
unstructured search e.g. using an intra-/internet search engine. If
there are multiple search results delivered, they have to be
applied to a similarity check. Thereby, e.g. by means of available
ontologies, related terms can be identified. For each most frequent
similar hit the similarity factor will be raised. In case of no
search hits for an item that is part of a taxonomy, the "father",
"grandfather", . . . relation can be searched instead of the term
for which the search failed. If there are search results on terms
inferred from taxonomies, this is documented and a reducing
taxonomy factor will be considered. Any search results entitled for
display may be condensed (technology available e.g. for smartphone
apps) or stripped such that they can be recognized within a moment.
The trustworthiness factor may get deducted by (multiplying with)
the similarity factor and by a taxonomy factor. The search results
are associated with the individual determined trustworthiness
factor and stored in the community thread glossary and linked to
the session transcript. Based on the updated community thread
glossary the tag cloud is recreated. Depending on a system-wide
policy based on the number of retrievals by the communication
thread community, the search result is also stored in the domain
glossary. Finally, the user interface is updated.
[0026] As a further option, the auto-corrected transcript can be
translated a specified language, e.g. to the standard language
defined by the company.
[0027] As further enhancements, sentiment detection technologies
can be applied to individual search results in order to apply a
weight factor for the trustworthiness factor or value, i.e. the
sentiment factors for: negative/ironic context, neutral context,
positive context. Proposed default values may be 0.1, 0.9, and 1,
respectively, for these contexts.
[0028] As a further option, search results on
structured/unstructured search are examined in respect to reader's
evaluation schemes and its grade may be used as another weight
factor for the trustworthiness factor, i.e. the community
evaluation factor: e.g. an evaluation "4 of 5 stars" results in a
weight factor of 0.8.
[0029] The trustworthiness factor may get further deducted by
(multiplying with) the sentiment factor and the community
evaluation factor.
[0030] As a further option, the user can judge the trustworthiness
of a selected item, e.g. using a move-over context menu and
override the value. A significant number of average override-values
from the community thread will be considered as an additional
weight when the item is stored in the domain glossary.
[0031] As described above, there is an interrelation between the
method and the system according to the invention. Therefore, it is
apparent that features described in connection with the method may
be present or even necessarily present also in the system, and vice
versa, although this may not be mentioned explicitly.
[0032] Other objects, features and advantages of the invention(s)
disclosed herein may become apparent from the following
description(s) thereof.
BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Reference will be made in detail to embodiments of the
disclosure, non-limiting examples of which may be illustrated in
the accompanying drawing figures (FIGS). The figures are generally
in the form of diagrams. Some elements in the figures may be
exaggerated, others may be omitted, for illustrative clarity. Some
figures may be in the form of diagrams. Although the invention is
generally described in the context of various exemplary
embodiments, it should be understood that it is not intended to
limit the invention to these particular embodiments, and individual
features of various embodiments may be combined with one another.
Any text (legends, notes, reference numerals and the like)
appearing on the drawings are incorporated by reference herein.
[0034] FIG. 1 is a diagram illustrating an exemplary speech
communication system which may be suitable for implementing various
embodiments of the invention.
[0035] FIG. 2 is a diagram showing a sequence of steps and events
which may occur or be present in an exemplary method of improving
communication in a speech communication system.
[0036] FIG. 3 is a diagram showing in more detail a sequence of
steps and events which may occur or be present in the step of
determination of the trustworthiness factor.
[0037] FIG. 4 is a diagram showing a sequence of steps and events
which may occur or be present in an exemplary method of similarity
checking as a part of the determination of the trustworthiness
factor.
[0038] FIG. 5 is a diagram showing a sequence of steps and events
which may occur or be present in an exemplary method of sentiment
detection which may be a part of determination of the
trustworthiness factor.
[0039] FIG. 6 is a diagram which shows in an exemplary manner some
components of the speech communication system, including a display,
on which a transcript window is shown and on which several keywords
together with their definition and the corresponding
trustworthiness factor are shown.
DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS
[0040] Various embodiments may be described to illustrate teachings
of the invention, and should be construed as illustrative rather
than limiting. It should be understood that it is not intended to
limit the invention to these particular embodiments. It should be
understood that some individual features of various embodiments may
be combined in different ways than shown, with one another. There
may be more than one invention described herein.
[0041] The embodiments and aspects thereof may be described and
illustrated in conjunction with systems, devices and methods which
are meant to be exemplary and illustrative, not limiting in scope.
Specific configurations and details may be set forth in order to
provide an understanding of the invention(s). However, it should be
apparent to one skilled in the art that the invention(s) may be
practiced without some of the specific details being presented
herein. Furthermore, some well-known steps or components may be
described only generally, or even omitted, for the sake of
illustrative clarity.
[0042] Reference herein to "one embodiment", "an embodiment", or
similar formulations, may mean that a particular feature,
structure, operation, or characteristic described in connection
with the embodiment, is included in at least one embodiment of the
present invention. Thus, the appearances of such phrases or
formulations herein are not necessarily all referring to the same
embodiment. Furthermore, various particular features, structures,
operations, or characteristics may be combined in any suitable
manner in one or more embodiments.
[0043] In the following descriptions, some specific details may be
set forth in order to provide an understanding of the invention(s)
disclosed herein. It should be apparent to those skilled in the art
that these invention(s) may be practiced without these specific
details. Headings (typically in uppercase letters) may be provided
as an aid to the reader, and should not be construed as
limiting.
[0044] Any dimensions and materials or processes set forth herein
should be considered to be approximate and exemplary, unless
otherwise indicated.
[0045] FIG. 1 illustrates an exemplary speech communication system
10 (also abbreviated "system 10") which comprises several
functional entities. These entities may be designed as distinct
units linked with each other or with a processing unit such as a
central processing unit ("CPU"), interconnected processors, a
microprocessor, or other type of processor, or they may be for
example tasks carried out by a CPU or other type of processor.
Furthermore, it is possible that these entities are a mixture of
these configurations. Item 12 indicates that the speech
communication system 10 is related to media conferencing. A media
recording unit 14 is part of the media conferencing 12 and is
connected with the speech communication system 10. A cloud 16
indicates that the speech communication system 10 is connected with
intranet and/or Internet. It should be appreciated that embodiments
of the communication system 10 may include a computer, a server, a
media conference server, a laptop computer, or a mobile computer
device such as a smart phone or internet appliance or tablet. The
system may include hardware elements such as a display (e.g. a
liquid crystal display, a touch actuatable display, a monitor),
non-transitory memory (e.g. a hard drive, flash memory, etc.), and
input devices (e.g. a keyboard, touch sensitive display screen,
mouse, scanner, reader, etc.) that are communicatively connected to
a processing unit (e.g. a central processing unit, at least one
processor, interconnected processors, etc.).
[0046] The speech communication system 10 may comprise a
transcripting unit 20, a spell-checker unit 22, a word highlighting
or word spotting unit 24, a semantic search unit 26, a similarity
checker unit 28, and an internet search unit 30. Furthermore, the
speech communication system 10 can comprise a marker unit 32 for
manual marking and a marking unit 34 for community marking of
certain words. Furthermore, a tag cloud creator 36 may be included.
A display 50 for displaying session tag clouds and a display 60 for
displaying session transcripts may be provided as well.
Furthermore, the system 10 includes a data storage 82 for a regular
dictionary (or several regular dictionaries), a data storage 84 for
a domain dictionary (which may also be called a glossary), a data
storage 86 for one or several communication threads created during
the speech communication, and a data storage 88 for storing a
directory of trusted information spaces. The system 10 may as well
include a data storage 90 for dealing with ontologies and
taxonomies. It goes without saying that in an alternative
embodiment at least some of the data storages 82-90 or other
entities mentioned before may be located outside the system 10 and
linked or connected with the system 10.
[0047] The specific functions of the above-mentioned functional
entities and their respective connections/connectivities as well as
the overall function of the speech communication system 10 may be
better understood from the explanation of an embodiment of the
method of improving communication in a speech communication system
10 as depicted in FIG. 2.
[0048] As shown in FIG. 2, in a step S101 a real-time media stream
or a corresponding recording is accessed to be processed in a step
S102, wherein either real-time recorded speech data or replayed
segments of speech data are used. In a following step S120 these
data are transcribed in order to create a transcribed portion. In
this example, the method is carried out sequentially on respective
segments of speech data. As an alternative, the method may be
carried out continuously on speech data as well. In a step S122 a
spell-check of the transcribed portion against a regular dictionary
stored in a data storage 82 is carried out in the spell-checker
unit 22. The spellchecked transcribed portion forms a so-called
communication thread.
[0049] The speech data are generated in this example in a telephone
call between a first user and a second user. It is clear that the
speech data may also stem from a telephone conference between more
than just two users or that these speech data result from a video
conference.
[0050] Thereafter, the spell-checker unit 22 carries out a
spell-check against a domain dictionary or domain glossary stored
in a data storage 84 which contains terms frequently used in a
certain domain or community. Thereafter, in a step S124 the
selecting unit or word spotting unit 24 spots words and terms which
may be keywords of the communication thread.
[0051] In a step S125, the found glossary items are linked to the
respective keywords. The steps S122, S123, S124, and S125 may be
regarded as a combined step S121.
[0052] After performing this combined step S121 for the first time,
the method gives a users the possibility to manually mark in a step
S132 words of the communication thread by using an input device in
the form of a manual marking unit 32 (which typically may be a
computer mouse) in order to indicate that the respective word is
regarded as a keyword. The fact that this possibility is open to
the user is indicated by the number "1" at the arrow pointing to
the manual marking step S132. After step S132, the combined step
S121 is carried out once again for the manually marked word or
words. After the conclusion of the combined step S121 for the
second time, the display on which the results of the steps carried
out so far are shown is updated in a step S110, as indicated by the
number "2" at the arrow pointing to step S110.
[0053] In a step S130, the internet search unit 30 performs a
structured and/or an unstructured intranet or internet search in
order to be able to produce at least one definition for each of the
keywords. It may happen that by step S130 just one "correct"
definition for a respective keyword is found or produced,
respectively, and it may well happen that several different
definitions are generated. In a step S126, the semantic search unit
26 performs a semantic intranet and/or internet search for refining
or correcting the search results of step 130. In a step 128
performed thereafter, the similarity checker unit 28 selects items
found on the basis of similarity. In a step S129, for the items
found, information is retrieved and stripped for unnecessary
portions.
[0054] In a step S400, the trustworthiness analysis unit 40
determines or calculates, respectively, a trustworthiness factor
(thereafter partly abbreviated with TWF) which indicates their
reliability/veracity/validity of the definition or definitions
generated so far. Then, in a step S131 the selected items are
linked to the spotted words and terms. In a step S133, the glossary
containing the communication thread is updated, i.e. the "new"
keywords are added to the communication thread glossary (which is
stored in the data storage 86). In a step S134, the manual markings
of the other users not yet considered in the description of the
invention so far are taken into account. In other words, the result
of the marking, selecting and determining of the TWF of those other
users is also taken into account. In a step S135, the cloud of the
session tags shown on the respective display unit 50 is updated. In
a step S137, the domain glossary stored in the data storage 84 is
updated. Finally, in a further step S110, the display showing the
information produced so far is updated again. At this point in
time, the method of the invention continues with the next real-time
capturing segment or with the next real-time replay segment, as the
case may be. It is of course also possible to apply this invention
to segments which are not recorded in real-time. This step may be
called a step S139.
[0055] In FIG. 3, details of the trustworthiness factor determining
step S400 are shown. In a step S402, the trustworthiness factor is
set to 100%, or 1, respectively. In a step S404, it is checked
whether the respective item was found in a dictionary or glossary.
If this is the case, the respective trustworthiness factor TWF'
from the domain dictionary is retrieved, and this TWF' replaces the
former TWF. Thereafter, in a step S490 this TWF is saved (in
connection with the respective item from the
dictionary/glossary).
[0056] In case the step S404 reveals that the respective item was
not found in a dictionary/glossary, in a step S414 it is checked
whether the items stem from trusted information spaces such as e.g.
Wikipedia. If this is the case, in a step S416 the TWF is
multiplied with a TWF'' which is associated with a (weighted)
average from the respective information spaces. Afterwards, in step
S500 a similarity check is performed, and then a sentiment
detection is carried out in a step S600. Details with respect to
the steps S500 and S600 may be found in connection with the
description of the FIGS. 4 and 5, respectively. Finally, in the
step S490 the calculated TWF is saved.
[0057] In case the step S414 reveals that the items are not from
trusted information spaces, in a step S424 it is checked whether
the items result from a structured or unstructured search. If this
is the case, the steps S500 of similarity checking and S600 of
sentiment detection are carried out, and in the subsequent step
S490 the respective TWF is saved.
[0058] If the step S424 reveals that the items are not from a
structured/unstructured search, in a step S426 it is checked
whether the items stem from a semantic search using ontologies. If
this is the case, again, the steps S500 of similarity checking and
S600 of sentiment detection are carried out, and finally the TWF is
saved in the step S490.
[0059] In case the step S426 reveals that the items are not from a
semantic search using ontologies, it is checked in a step S428
whether the items are from a semantic search using taxonomies. If
the answer to this check is "no", the respective TWF is saved in
the step S490. In case the answer to this check is "yes", the
similarity checking step S500 is carried out. Thereafter, in a step
S430 the present TWF is multiplied with a taxonomy factor. In other
words, in case there are no search hits for an item that is part of
a taxonomy, the "father", "grandfather", relation can be searched
instead of the term for which the search failed. If there are
search results on terms inferred from taxonomies, this is
documented and a reducing taxonomy factor will be applied to the
TWF. This taxonomy factor may result in a reduction of the TWF. The
taxonomy factor represents the "distance" in the hierarchy between
two terms, and it is e.g. for the father 75%, for the grandfather
50%, for the grand-grandfather 25%, and for the brother 85%. After
that, the sentiment detection step S600 is carried out, and the
respective TWF is saved in the step S490.
[0060] In FIG. 4 the process of similarity checking, which was
summarized as one single step S500 in the previous discussion, is
explained in detail. In a step S502, the first item found is
latched in a buffer. In a subsequent step S504 it is checked,
whether there are further similar items. In case there are further
similar items, in a step S506 the next item found is latched in a
buffer and compared in a step S508 with the previous item in order
to perform a similarity checking. To give an example, SML
(=Semantic Measures Library) and a respective toolkit can be used
to compute semantic similarity between semantic elements/terms. The
steps S506 and S508 are carried out for each further item found. As
soon as there are no further items, in a step S510 the item list is
reduced to those items with the most frequent similarity. In a
subsequent step S512 a similarity factor is calculated which is the
number of the most frequent similar items divided by the total
number of items. In a subsequent step S514 the current TWF is
multiplied with the similarity factor calculated in step S512. In a
step S516 this modified TWF (which is the most current TWF) is
returned.
[0061] FIG. 5 explains how the sentiment detection step S600 is
carried out, which was summarized as one single step S600 in the
discussion above. In a step S602 a sentiment analysis is performed
in order to find out whether in the definition or definitions found
so far for the keywords any sentiment is included. Examples for the
sentiment may be negative or ironic, positive, or neutral. In other
words, the sentiment detection reveals whether one of the users has
a personal assessment or appreciation of a certain word or term
which is manifested in the respective communication thread. In case
the sentiment analysis in step S602 reveals that there is a neutral
sentiment, the current TWF is multiplied with a "neutral" sentiment
factor in a step S610.
[0062] Afterwards, in a step S650, a check for community evaluation
is performed. If it is found that the community, i.e. other
readers, have given an evaluation, a community evaluation factor is
calculated and multiplied with the TWF found so far. If for example
the community gives a ranking of 80% for a certain definition of a
keyword, the community evaluation factor, which is a weight factor,
would result in a number of 0.8. This calculation and
multiplication is carried out in a step S652. Afterwards, the
modified TWF is returned in a step S660. In case no community
evaluation can be found, the modified TWF is returned "directly"
without any multiplication with a community evaluation factor in
step S660.
[0063] In case the sentiment analysis in step S602 reveals that
there is a positive sentiment, the current TWF is multiplied with a
"positive" sentiment factor in a step S620. Then, the steps
S650-S660 are carried out. If, however, the sentiment analysis in
step S602 reveals that there is a negative or ironic sentiment, the
current TWF is multiplied with a "negative" sentiment factor in a
step S630. Then, the steps S650-S660 are carried out. In case in
the sentiment analysis in step S602 no sentiment at all is found,
the steps S650-S660 are carried out without any multiplication of
the TWF with a sentiment factor.
[0064] Just to give an example, the sentiment factor may have a
value of 0.1 for a negative or an ironic, a value of 0.9 for a
neutral, and a value of 1 for a positive sentiment.
[0065] Finally, the respective displays (e.g. user interfaces shown
via a display unit such as a liquid crystal display or monitor),
like the user interface 70, are updated. That means that a series
of updated information is displayed via a display unit. One example
for this display is given in FIG. 6. This view illustrates the way
a user experiences the present speech communication system 10 and
the corresponding method carried out by the system 10. On a
transcript window corresponding to a user interface 70, a tag cloud
is displayed which is schematically and generally referenced with
the numeral 51. Three specific tags are shown here as random
examples: The tag 52 refers to "big data", the tag 53 refers to
"semantic web", and the tag 54 is directed to "augmented reality".
In the respective "stars" in the tags 52 to 54, the corresponding
trustworthiness factors are displayed. In other words, the keyword
"big data" has a TWF of 95, the keyword "semantic web" has a TWF of
98, and "augmented reality" has a TWF of only 85. One portion of
the display is a conference control graphical user interface which
is referenced with the numeral 11. It goes without saying that the
display may be updated at any step or at any point in time
considered to be useful for the process.
[0066] It may well be contemplated to give certain privileges to
specified users, e.g. different user rights, in order to allow for
overriding the trustworthiness factor.
[0067] While the invention(s) has/have been described with respect
to a limited number of embodiments, these should not be construed
as limitations on the scope of the invention(s), but rather as
examples of some of the embodiments. Those skilled in the art may
envision other possible variations, modifications, and
implementations that are also within the scope of the invention(s),
based on the disclosure(s) set forth herein.
* * * * *