SlideShare ist ein Scribd-Unternehmen logo
1 von 6
Analyzing Miscommunications.
             Ravi Kiran Holur Vijay           Peter Thermos         Henning Schulzrinne

                     Computer Science Department, Columbia University, NY.

Abstract.                                              Also, efforts were made to produce a task-
                                                       oriented speech corpus in a multi-cultural and
We explore some metrics that might possibly            multi-speaker setting. For this, we used the
indicate       occurrence      of       possible       game Counter Strike[6] to simulate a virtual
miscommunication during a task-oriented                world where players could communicate. The
conversation. The metrics explored are                 players involved in this exercise were from
motivated by real-world observations and cover         diverse cultural backgrounds. Also, English was
lexical, structural and semantic attributes. We        not the native language in most of the cases.
develop a test bed for evaluating the metrics
and execute the metrics on a task-oriented             2. Metrics.
corpus. Also, efforts were made to produce a
task-oriented speech corpus in a multi-speaker         2.1 Lexical level.
and multi-cultural setting.
                                                       Word ambiguity induced by context of use can
                                                       happen when a word can have multiple
1. Introduction.
                                                       interpretations based on the context in which it
If we imagine ourselves as participants in a task-     is used. Here, the required context is provided
oriented conversation, we can see that                 by the sentence in which the word appears. For
miscommunications indeed do happen [1].                quantifying this ambiguity, we use two metrics
Further,     if     we     think     about     the     – Score output of WSD algorithms and SD of
miscommunication detection and repair process          sense frequencies from WordNet (SensesSD) [7].
adopted by humans, we can observe that it              For the first metric, we try out a couple of WSD
occurs at various logical levels - word, sentence      algorithms [8] [9]. Both of these algorithms take a
and dialog. Based on some real world                   context and a target word as input and give the
observations and previous works [1][2][3][4], we       most probable sense for the target word as
will explore some metrics corresponding to             output, using WordNet as the dictionary. But,
lexical, syntactic and semantic levels. At lexical     we need a score using which we could
level, we can calculate a metric that could be         determine how probable the best probable
used to quantify the ambiguity of each word            sense was. For this purpose, we take the score
(both dependent and independent of context).           as the number of overlaps of the sense with
At the structural level, we can calculate a metric     maximum overlaps and normalize it. For the
based on syntactic priming [3][4]. At the semantic     second metric, we calculate the standard
level, we can calculate a metric based on the          deviation of the frequencies of all the possible
sentimental polarity (positive or negative) of         senses of a given word, using WordNet. After
the sentence. In order to evaluate the metrics,        implementing and executing the above metrics
we run the implementations of these on a task-         on a task-oriented corpus, the following
oriented corpus [5].                                   observations were made. If the score is less
than 0 or more than 1 (like really, sorry, nice,              category, with low scores (less than .15) and
left, stop, right, back, there etc.), we can see              high SensesSD (greater than 50). So, basically
that the words might lead to potential                        we need to identify words with low scores and
misunderstanding more readily. Now, if the                    high SensesSD, which might give us an
score is between 0 and 1, we need to look at                  approximate measure for the misunderstanding
SensesSD to decide if the word might lead to                  it can cause in the given context (sentence). A
potential misunderstanding. Some words (here,                 couple of graphs of Scores vs. No. of words is
there, work, fire, want, think etc.) are in this              shown in [Figure 1] & [Figure 2].




                       Figure 1 - Distribution of Scores for WordNet Lesk WSD algorithm - I




                       Figure 2 - Distribution of Scores for WordNet Lesk WSD algorithm - II
2.2 Structural level.                                           The implementation was executed on the Map
                                                                Task Corpus [5]. The task success metric used in
At the sentence level, we explore a metric                      Map Task corpus was the attribute to be
based on Syntax priming [3][4]. We can use the                  predicted by the regression model. The
metric to predict the task outcome using a                      performance of the evaluation is shown in
regression model (SVM based regression) and                     [Figure 4], the average error rate being -
thus evaluate how effective it is for predicting                12.7622217.
task success. The architecture is as shown in
[Figure 3].




                        Figure 3 - Architecture for implementation of Structural level metrics.
Figure 4 - Performance of Structural Priming based metrics.




2.3 Semantic level.                                         corpus, the results obtained are as shown in
                                                            [Figure 5].
At the semantic level, we explore a metric
related to sentiment (positive, negative and                The correlation coefficients between Task
neutral) of the sentence. We define the                     metric and No. of Positive, Negative and Neutral
sentiment of a sentence as the sum of the                   sentences were -0.07, -0.007 and -0.1023
sentiment of the adjectives that the sentence               respectively.
contains. The sentiments of the adjectives were
inferred using SentiWordNet[11]. When the
implementation was executed on the Map Task




                                 Figure 5 - Effect of Sentiment on Task Metric.
3. Task-oriented spoken dialog                              teams of human players played against each
                                                            other and where teams of human players
corpus.
                                                            played against the computer bots. The in-game
In order to continue further explorations of the            conversations among players and the final
problem, efforts were made to produce a task-               scores were recorded for each of the games.
oriented spoken dialog corpus in a multi-                   The next step would be to transcribe the audios
cultural setting. For this, we used the popular             of each of the games into text, either manually
FPS “Counter Strike” [6]. There were two teams,             or in a semi-supervised manner. A couple of
with approximately two to four players on each              screenshots from the game, along with the
team. Each session consisted of games where                 metrics used to measure the team’s success is
                                                            shown in [Figures 6, 7].




                                Figure 6 - In-Game world and Success Metrics I




                                Figure 7 - In-Game world and Success Metrics II
4. References.
[1] Words Are Mightier Than Swords … and Yet Miscommunication Costs Lives! - Stephen Poteet, Jitu
Patel, Cheryl Giammanco.

[2] Fatal Words: Communication Clashes and Aircraft Crashes - Steven Cushing.

[3] TOWARD A MECHANISTIC PSYCHOLOGY OF DIALOGUE - Martin J. Pickering, Simon Garrod.

[4] Priming of Syntactic Rules in Task-Oriented Dialogue and Spontaneous Conversation - David Reitter,
Johanna D. Moore, Frank Keller.

[5] Anderson, A., Bader, M., Bard, E., Boyle, E., Doherty, G. M., Garrod, S., Isard, S., Kowtko, J.,
McAllister, J., Miller, J., Sotillo, C., Thompson, H. S. and Weinert, R. (1991). The HCRC Map Task Corpus.
Language and Speech, 34, pp. 351-366.

[6] Counter Strike - http://store.steampowered.com/app/240/

[7] C. Fellbaum, editor. WordNet: An electronic lexical database. MITPress, 1998.

[8] Evaluating variants of the Lesk Approach for Disambiguating Words” by Florentina Vasilescu, Philippe
Langlais and Guy Lapalme.

[9] Adapting the Lesk Algorithm for Word Sense Disambiguation to WordNet by Satanjeev Banerjee.

[10] Stanford Parser - http://nlp.stanford.edu/software/lex-parser.shtml

[11] Andrea Esuli and Fabrizio Sebastiani. SentiWordNet: A Publicly Available Lexical Resource for
Opinion Mining. In Proceedings of LREC-06, 5th Conference on Language Resources and Evaluation,
Genova, IT, 2006, pp. 417-422.

[12] Ian H. Witten and Eibe Frank (2005) "Data Mining: Practical machine learning tools and techniques",
2nd Edition, Morgan Kaufmann, San Francisco, 2005.

Weitere ähnliche Inhalte

Was ist angesagt?

security framework
security frameworksecurity framework
security frameworkJihad Labban
 
Report for Speech Emotion Recognition
Report for Speech Emotion RecognitionReport for Speech Emotion Recognition
Report for Speech Emotion RecognitionDongang (Sean) Wang
 
IRJET- Emotion recognition using Speech Signal: A Review
IRJET-  	  Emotion recognition using Speech Signal: A ReviewIRJET-  	  Emotion recognition using Speech Signal: A Review
IRJET- Emotion recognition using Speech Signal: A ReviewIRJET Journal
 
ASERS-LSTM: Arabic Speech Emotion Recognition System Based on LSTM Model
ASERS-LSTM: Arabic Speech Emotion Recognition System Based on LSTM ModelASERS-LSTM: Arabic Speech Emotion Recognition System Based on LSTM Model
ASERS-LSTM: Arabic Speech Emotion Recognition System Based on LSTM Modelsipij
 
Learning analytics to identify exploratory dialogue in online discussions
Learning analytics to identify exploratory dialogue in online discussionsLearning analytics to identify exploratory dialogue in online discussions
Learning analytics to identify exploratory dialogue in online discussionsRebecca Ferguson
 
Automatic speech emotion and speaker recognition based on hybrid gmm and ffbnn
Automatic speech emotion and speaker recognition based on hybrid gmm and ffbnnAutomatic speech emotion and speaker recognition based on hybrid gmm and ffbnn
Automatic speech emotion and speaker recognition based on hybrid gmm and ffbnnijcsa
 
ASERS-CNN: ARABIC SPEECH EMOTION RECOGNITION SYSTEM BASED ON CNN MODEL
ASERS-CNN: ARABIC SPEECH EMOTION RECOGNITION SYSTEM BASED ON CNN MODELASERS-CNN: ARABIC SPEECH EMOTION RECOGNITION SYSTEM BASED ON CNN MODEL
ASERS-CNN: ARABIC SPEECH EMOTION RECOGNITION SYSTEM BASED ON CNN MODELsipij
 
2.7 article dec 2016 edition register journal vol 9 no 2
2.7 article   dec 2016 edition register journal vol 9 no 22.7 article   dec 2016 edition register journal vol 9 no 2
2.7 article dec 2016 edition register journal vol 9 no 2Faisal Pak
 

Was ist angesagt? (10)

Jw2417001703
Jw2417001703Jw2417001703
Jw2417001703
 
security framework
security frameworksecurity framework
security framework
 
Report for Speech Emotion Recognition
Report for Speech Emotion RecognitionReport for Speech Emotion Recognition
Report for Speech Emotion Recognition
 
IRJET- Emotion recognition using Speech Signal: A Review
IRJET-  	  Emotion recognition using Speech Signal: A ReviewIRJET-  	  Emotion recognition using Speech Signal: A Review
IRJET- Emotion recognition using Speech Signal: A Review
 
ASERS-LSTM: Arabic Speech Emotion Recognition System Based on LSTM Model
ASERS-LSTM: Arabic Speech Emotion Recognition System Based on LSTM ModelASERS-LSTM: Arabic Speech Emotion Recognition System Based on LSTM Model
ASERS-LSTM: Arabic Speech Emotion Recognition System Based on LSTM Model
 
Learning analytics to identify exploratory dialogue in online discussions
Learning analytics to identify exploratory dialogue in online discussionsLearning analytics to identify exploratory dialogue in online discussions
Learning analytics to identify exploratory dialogue in online discussions
 
Automatic speech emotion and speaker recognition based on hybrid gmm and ffbnn
Automatic speech emotion and speaker recognition based on hybrid gmm and ffbnnAutomatic speech emotion and speaker recognition based on hybrid gmm and ffbnn
Automatic speech emotion and speaker recognition based on hybrid gmm and ffbnn
 
ASERS-CNN: ARABIC SPEECH EMOTION RECOGNITION SYSTEM BASED ON CNN MODEL
ASERS-CNN: ARABIC SPEECH EMOTION RECOGNITION SYSTEM BASED ON CNN MODELASERS-CNN: ARABIC SPEECH EMOTION RECOGNITION SYSTEM BASED ON CNN MODEL
ASERS-CNN: ARABIC SPEECH EMOTION RECOGNITION SYSTEM BASED ON CNN MODEL
 
call for papers, research paper publishing, where to publish research paper, ...
call for papers, research paper publishing, where to publish research paper, ...call for papers, research paper publishing, where to publish research paper, ...
call for papers, research paper publishing, where to publish research paper, ...
 
2.7 article dec 2016 edition register journal vol 9 no 2
2.7 article   dec 2016 edition register journal vol 9 no 22.7 article   dec 2016 edition register journal vol 9 no 2
2.7 article dec 2016 edition register journal vol 9 no 2
 

Ähnlich wie Analyzing Miscommunication

An Improved Approach for Word Ambiguity Removal
An Improved Approach for Word Ambiguity RemovalAn Improved Approach for Word Ambiguity Removal
An Improved Approach for Word Ambiguity RemovalWaqas Tariq
 
Analysis of Opinionated Text for Opinion Mining
Analysis of Opinionated Text for Opinion MiningAnalysis of Opinionated Text for Opinion Mining
Analysis of Opinionated Text for Opinion Miningmlaij
 
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIESTHE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIESkevig
 
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIESTHE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIESkevig
 
Doc format.
Doc format.Doc format.
Doc format.butest
 
OPTIMIZATION OF CROSS DOMAIN SENTIMENT ANALYSIS USING SENTIWORDNET
OPTIMIZATION OF CROSS DOMAIN SENTIMENT ANALYSIS USING SENTIWORDNETOPTIMIZATION OF CROSS DOMAIN SENTIMENT ANALYSIS USING SENTIWORDNET
OPTIMIZATION OF CROSS DOMAIN SENTIMENT ANALYSIS USING SENTIWORDNETijfcstjournal
 
Generating domain specific sentiment lexicons using the Web Directory
Generating domain specific sentiment lexicons using the Web Directory Generating domain specific sentiment lexicons using the Web Directory
Generating domain specific sentiment lexicons using the Web Directory acijjournal
 
DataChat_FinalPaper
DataChat_FinalPaperDataChat_FinalPaper
DataChat_FinalPaperUrjit Patel
 
Welcome to International Journal of Engineering Research and Development (IJERD)
Welcome to International Journal of Engineering Research and Development (IJERD)Welcome to International Journal of Engineering Research and Development (IJERD)
Welcome to International Journal of Engineering Research and Development (IJERD)IJERD Editor
 
Natural Language Processing
Natural Language ProcessingNatural Language Processing
Natural Language Processingpunedevscom
 
NBLex: emotion prediction in Kannada-English code-switchtext using naïve baye...
NBLex: emotion prediction in Kannada-English code-switchtext using naïve baye...NBLex: emotion prediction in Kannada-English code-switchtext using naïve baye...
NBLex: emotion prediction in Kannada-English code-switchtext using naïve baye...IJECEIAES
 
Chunker Based Sentiment Analysis and Tense Classification for Nepali Text
Chunker Based Sentiment Analysis and Tense Classification for Nepali TextChunker Based Sentiment Analysis and Tense Classification for Nepali Text
Chunker Based Sentiment Analysis and Tense Classification for Nepali Textkevig
 
Chunker Based Sentiment Analysis and Tense Classification for Nepali Text
Chunker Based Sentiment Analysis and Tense Classification for Nepali TextChunker Based Sentiment Analysis and Tense Classification for Nepali Text
Chunker Based Sentiment Analysis and Tense Classification for Nepali Textkevig
 
Evaluating sentiment analysis and word embedding techniques on Brexit
Evaluating sentiment analysis and word embedding techniques on BrexitEvaluating sentiment analysis and word embedding techniques on Brexit
Evaluating sentiment analysis and word embedding techniques on BrexitIAESIJAI
 
Speech emotion recognition using 2D-convolutional neural network
Speech emotion recognition using 2D-convolutional neural  networkSpeech emotion recognition using 2D-convolutional neural  network
Speech emotion recognition using 2D-convolutional neural networkIJECEIAES
 
Trust evaluation using an improved context
Trust evaluation using an improved contextTrust evaluation using an improved context
Trust evaluation using an improved contextijbiss
 
Trust Evaluation Using an Improved Context Similarity Measurement
Trust Evaluation Using an Improved Context Similarity MeasurementTrust Evaluation Using an Improved Context Similarity Measurement
Trust Evaluation Using an Improved Context Similarity Measurementijbiss
 

Ähnlich wie Analyzing Miscommunication (20)

228-SE3001_2
228-SE3001_2228-SE3001_2
228-SE3001_2
 
An Improved Approach for Word Ambiguity Removal
An Improved Approach for Word Ambiguity RemovalAn Improved Approach for Word Ambiguity Removal
An Improved Approach for Word Ambiguity Removal
 
Analysis of Opinionated Text for Opinion Mining
Analysis of Opinionated Text for Opinion MiningAnalysis of Opinionated Text for Opinion Mining
Analysis of Opinionated Text for Opinion Mining
 
Bi-lingual Word Sense Induction
Bi-lingual Word Sense InductionBi-lingual Word Sense Induction
Bi-lingual Word Sense Induction
 
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIESTHE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
 
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIESTHE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
THE ABILITY OF WORD EMBEDDINGS TO CAPTURE WORD SIMILARITIES
 
Doc format.
Doc format.Doc format.
Doc format.
 
OPTIMIZATION OF CROSS DOMAIN SENTIMENT ANALYSIS USING SENTIWORDNET
OPTIMIZATION OF CROSS DOMAIN SENTIMENT ANALYSIS USING SENTIWORDNETOPTIMIZATION OF CROSS DOMAIN SENTIMENT ANALYSIS USING SENTIWORDNET
OPTIMIZATION OF CROSS DOMAIN SENTIMENT ANALYSIS USING SENTIWORDNET
 
Analyse de sentiment et classification par approche neuronale en Python et Weka
Analyse de sentiment et classification par approche neuronale en Python et WekaAnalyse de sentiment et classification par approche neuronale en Python et Weka
Analyse de sentiment et classification par approche neuronale en Python et Weka
 
Generating domain specific sentiment lexicons using the Web Directory
Generating domain specific sentiment lexicons using the Web Directory Generating domain specific sentiment lexicons using the Web Directory
Generating domain specific sentiment lexicons using the Web Directory
 
DataChat_FinalPaper
DataChat_FinalPaperDataChat_FinalPaper
DataChat_FinalPaper
 
Welcome to International Journal of Engineering Research and Development (IJERD)
Welcome to International Journal of Engineering Research and Development (IJERD)Welcome to International Journal of Engineering Research and Development (IJERD)
Welcome to International Journal of Engineering Research and Development (IJERD)
 
Natural Language Processing
Natural Language ProcessingNatural Language Processing
Natural Language Processing
 
NBLex: emotion prediction in Kannada-English code-switchtext using naïve baye...
NBLex: emotion prediction in Kannada-English code-switchtext using naïve baye...NBLex: emotion prediction in Kannada-English code-switchtext using naïve baye...
NBLex: emotion prediction in Kannada-English code-switchtext using naïve baye...
 
Chunker Based Sentiment Analysis and Tense Classification for Nepali Text
Chunker Based Sentiment Analysis and Tense Classification for Nepali TextChunker Based Sentiment Analysis and Tense Classification for Nepali Text
Chunker Based Sentiment Analysis and Tense Classification for Nepali Text
 
Chunker Based Sentiment Analysis and Tense Classification for Nepali Text
Chunker Based Sentiment Analysis and Tense Classification for Nepali TextChunker Based Sentiment Analysis and Tense Classification for Nepali Text
Chunker Based Sentiment Analysis and Tense Classification for Nepali Text
 
Evaluating sentiment analysis and word embedding techniques on Brexit
Evaluating sentiment analysis and word embedding techniques on BrexitEvaluating sentiment analysis and word embedding techniques on Brexit
Evaluating sentiment analysis and word embedding techniques on Brexit
 
Speech emotion recognition using 2D-convolutional neural network
Speech emotion recognition using 2D-convolutional neural  networkSpeech emotion recognition using 2D-convolutional neural  network
Speech emotion recognition using 2D-convolutional neural network
 
Trust evaluation using an improved context
Trust evaluation using an improved contextTrust evaluation using an improved context
Trust evaluation using an improved context
 
Trust Evaluation Using an Improved Context Similarity Measurement
Trust Evaluation Using an Improved Context Similarity MeasurementTrust Evaluation Using an Improved Context Similarity Measurement
Trust Evaluation Using an Improved Context Similarity Measurement
 

Kürzlich hochgeladen

SQL Database Design For Developers at php[tek] 2024
SQL Database Design For Developers at php[tek] 2024SQL Database Design For Developers at php[tek] 2024
SQL Database Design For Developers at php[tek] 2024Scott Keck-Warren
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreternaman860154
 
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 3652toLead Limited
 
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024BookNet Canada
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024Rafal Los
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationMichael W. Hawkins
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Miguel Araújo
 
Slack Application Development 101 Slides
Slack Application Development 101 SlidesSlack Application Development 101 Slides
Slack Application Development 101 Slidespraypatel2
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitecturePixlogix Infotech
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024The Digital Insurer
 
Google AI Hackathon: LLM based Evaluator for RAG
Google AI Hackathon: LLM based Evaluator for RAGGoogle AI Hackathon: LLM based Evaluator for RAG
Google AI Hackathon: LLM based Evaluator for RAGSujit Pal
 
Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Allon Mureinik
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersThousandEyes
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptxHampshireHUG
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationRadu Cotescu
 
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationSafe Software
 
🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘RTylerCroy
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesSinan KOZAK
 

Kürzlich hochgeladen (20)

SQL Database Design For Developers at php[tek] 2024
SQL Database Design For Developers at php[tek] 2024SQL Database Design For Developers at php[tek] 2024
SQL Database Design For Developers at php[tek] 2024
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
 
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
 
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day Presentation
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
 
Slack Application Development 101 Slides
Slack Application Development 101 SlidesSlack Application Development 101 Slides
Slack Application Development 101 Slides
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC Architecture
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
Google AI Hackathon: LLM based Evaluator for RAG
Google AI Hackathon: LLM based Evaluator for RAGGoogle AI Hackathon: LLM based Evaluator for RAG
Google AI Hackathon: LLM based Evaluator for RAG
 
Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organization
 
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen Frames
 

Analyzing Miscommunication

  • 1. Analyzing Miscommunications. Ravi Kiran Holur Vijay Peter Thermos Henning Schulzrinne Computer Science Department, Columbia University, NY. Abstract. Also, efforts were made to produce a task- oriented speech corpus in a multi-cultural and We explore some metrics that might possibly multi-speaker setting. For this, we used the indicate occurrence of possible game Counter Strike[6] to simulate a virtual miscommunication during a task-oriented world where players could communicate. The conversation. The metrics explored are players involved in this exercise were from motivated by real-world observations and cover diverse cultural backgrounds. Also, English was lexical, structural and semantic attributes. We not the native language in most of the cases. develop a test bed for evaluating the metrics and execute the metrics on a task-oriented 2. Metrics. corpus. Also, efforts were made to produce a task-oriented speech corpus in a multi-speaker 2.1 Lexical level. and multi-cultural setting. Word ambiguity induced by context of use can happen when a word can have multiple 1. Introduction. interpretations based on the context in which it If we imagine ourselves as participants in a task- is used. Here, the required context is provided oriented conversation, we can see that by the sentence in which the word appears. For miscommunications indeed do happen [1]. quantifying this ambiguity, we use two metrics Further, if we think about the – Score output of WSD algorithms and SD of miscommunication detection and repair process sense frequencies from WordNet (SensesSD) [7]. adopted by humans, we can observe that it For the first metric, we try out a couple of WSD occurs at various logical levels - word, sentence algorithms [8] [9]. Both of these algorithms take a and dialog. Based on some real world context and a target word as input and give the observations and previous works [1][2][3][4], we most probable sense for the target word as will explore some metrics corresponding to output, using WordNet as the dictionary. But, lexical, syntactic and semantic levels. At lexical we need a score using which we could level, we can calculate a metric that could be determine how probable the best probable used to quantify the ambiguity of each word sense was. For this purpose, we take the score (both dependent and independent of context). as the number of overlaps of the sense with At the structural level, we can calculate a metric maximum overlaps and normalize it. For the based on syntactic priming [3][4]. At the semantic second metric, we calculate the standard level, we can calculate a metric based on the deviation of the frequencies of all the possible sentimental polarity (positive or negative) of senses of a given word, using WordNet. After the sentence. In order to evaluate the metrics, implementing and executing the above metrics we run the implementations of these on a task- on a task-oriented corpus, the following oriented corpus [5]. observations were made. If the score is less
  • 2. than 0 or more than 1 (like really, sorry, nice, category, with low scores (less than .15) and left, stop, right, back, there etc.), we can see high SensesSD (greater than 50). So, basically that the words might lead to potential we need to identify words with low scores and misunderstanding more readily. Now, if the high SensesSD, which might give us an score is between 0 and 1, we need to look at approximate measure for the misunderstanding SensesSD to decide if the word might lead to it can cause in the given context (sentence). A potential misunderstanding. Some words (here, couple of graphs of Scores vs. No. of words is there, work, fire, want, think etc.) are in this shown in [Figure 1] & [Figure 2]. Figure 1 - Distribution of Scores for WordNet Lesk WSD algorithm - I Figure 2 - Distribution of Scores for WordNet Lesk WSD algorithm - II
  • 3. 2.2 Structural level. The implementation was executed on the Map Task Corpus [5]. The task success metric used in At the sentence level, we explore a metric Map Task corpus was the attribute to be based on Syntax priming [3][4]. We can use the predicted by the regression model. The metric to predict the task outcome using a performance of the evaluation is shown in regression model (SVM based regression) and [Figure 4], the average error rate being - thus evaluate how effective it is for predicting 12.7622217. task success. The architecture is as shown in [Figure 3]. Figure 3 - Architecture for implementation of Structural level metrics.
  • 4. Figure 4 - Performance of Structural Priming based metrics. 2.3 Semantic level. corpus, the results obtained are as shown in [Figure 5]. At the semantic level, we explore a metric related to sentiment (positive, negative and The correlation coefficients between Task neutral) of the sentence. We define the metric and No. of Positive, Negative and Neutral sentiment of a sentence as the sum of the sentences were -0.07, -0.007 and -0.1023 sentiment of the adjectives that the sentence respectively. contains. The sentiments of the adjectives were inferred using SentiWordNet[11]. When the implementation was executed on the Map Task Figure 5 - Effect of Sentiment on Task Metric.
  • 5. 3. Task-oriented spoken dialog teams of human players played against each other and where teams of human players corpus. played against the computer bots. The in-game In order to continue further explorations of the conversations among players and the final problem, efforts were made to produce a task- scores were recorded for each of the games. oriented spoken dialog corpus in a multi- The next step would be to transcribe the audios cultural setting. For this, we used the popular of each of the games into text, either manually FPS “Counter Strike” [6]. There were two teams, or in a semi-supervised manner. A couple of with approximately two to four players on each screenshots from the game, along with the team. Each session consisted of games where metrics used to measure the team’s success is shown in [Figures 6, 7]. Figure 6 - In-Game world and Success Metrics I Figure 7 - In-Game world and Success Metrics II
  • 6. 4. References. [1] Words Are Mightier Than Swords … and Yet Miscommunication Costs Lives! - Stephen Poteet, Jitu Patel, Cheryl Giammanco. [2] Fatal Words: Communication Clashes and Aircraft Crashes - Steven Cushing. [3] TOWARD A MECHANISTIC PSYCHOLOGY OF DIALOGUE - Martin J. Pickering, Simon Garrod. [4] Priming of Syntactic Rules in Task-Oriented Dialogue and Spontaneous Conversation - David Reitter, Johanna D. Moore, Frank Keller. [5] Anderson, A., Bader, M., Bard, E., Boyle, E., Doherty, G. M., Garrod, S., Isard, S., Kowtko, J., McAllister, J., Miller, J., Sotillo, C., Thompson, H. S. and Weinert, R. (1991). The HCRC Map Task Corpus. Language and Speech, 34, pp. 351-366. [6] Counter Strike - http://store.steampowered.com/app/240/ [7] C. Fellbaum, editor. WordNet: An electronic lexical database. MITPress, 1998. [8] Evaluating variants of the Lesk Approach for Disambiguating Words” by Florentina Vasilescu, Philippe Langlais and Guy Lapalme. [9] Adapting the Lesk Algorithm for Word Sense Disambiguation to WordNet by Satanjeev Banerjee. [10] Stanford Parser - http://nlp.stanford.edu/software/lex-parser.shtml [11] Andrea Esuli and Fabrizio Sebastiani. SentiWordNet: A Publicly Available Lexical Resource for Opinion Mining. In Proceedings of LREC-06, 5th Conference on Language Resources and Evaluation, Genova, IT, 2006, pp. 417-422. [12] Ian H. Witten and Eibe Frank (2005) "Data Mining: Practical machine learning tools and techniques", 2nd Edition, Morgan Kaufmann, San Francisco, 2005.