Deja Bond

Twitter is a powerful social platform that continues to generate tremendous amounts of data on a wide range of topics such as politics. Previous mining of Twitter data has resulted in models that could use a users network to predict geolocation; and to predict an individual’s political ideology, separately. In this paper, we sought to create a model that only utilizes a user’s Twitter network (followers and accounts followed) to predict an individual’s political leanings and… Show more

Twitter is a powerful social platform that continues to generate tremendous amounts of data on a wide range of topics such as politics. Previous mining of Twitter data has resulted in models that could use a users network to predict geolocation; and to predict an individual’s political ideology, separately. In this paper, we sought to create a model that only utilizes a user’s Twitter network (followers and accounts followed) to predict an individual’s political leanings and geolocation synchronously. Using network embeddings as input we reduced incredibly large and sparse matrices to a more meaningful and manageable dimension after-which we implemented models to predict outcomes. We implemented ModularizedNon-Negative Matrix Factorization(M-NMF) and compared it to Large-scale Information Network Embedding (LINE), to create the embedding and the output was the embedding for the network, for which the output embedding was predictive on political ideology and geolocation. Random Forestregressor was implemented as the baseline learning model. After which we then used linear regression to understand linear relationships. Understanding Twitter’s network at this level could help to understand the policy preferences of individuals and as such political entities could gain better insight into trends and issues important to geographically disparate, but ideologically alike electorate groups. Show less

Speech is a way of communication among humans and is made up of two parts, namely sound and sense. Speech Recognition is the process in which certain words of a particular speaker are automatically recognized, based on the information included in individual speech waves. The sound can be visualized and analyzed in several ways, the recorded signal which is called test data is compared with the original signal which is called the trained data. Speech recognition is used in language learning… Show more

Speech is a way of communication among humans and is made up of two parts, namely sound and sense. Speech Recognition is the process in which certain words of a particular speaker are automatically recognized, based on the information included in individual speech waves. The sound can be visualized and analyzed in several ways, the recorded signal which is called test data is compared with the original signal which is called the trained data. Speech recognition is used in language learning, home automation, virtual assistance, medical documentation, fighter aircraft command and control, dictation, transcription of recorded speech, searching audio documents and interactive spoken dialogue among other things. At the core of all speech recognition systems are statistical models representing the sounds of the language to be determined. Speech is thus modeled as a sequence of spectral vectors spanning audio frequency range, and the hidden Markov model (HMM) provides a framework for constructing such models, it is the basic model of speech recognition. Hidden Markov Model (HMM) is a generative probabilistic model in which a sequence of observable variables is generated by a sequence of the hidden states. Transitions between hidden states are assumed to have the form of a (first-order) Markov chain and can be specified by the start probability vector, and a transition probability matrix. The reason it is called a Hidden Markov Model is that we are constructing an inference model based on the assumptions of a Markov process. The Markov process assumption is simply that: the future is independent of the past given the present Show less