Difference between revisions of "Probabilistic Latent Semantic Analysis (PLSA)"
m |
m |
||
| Line 13: | Line 13: | ||
* [[Term Frequency–Inverse Document Frequency (TF-IDF)]] | * [[Term Frequency–Inverse Document Frequency (TF-IDF)]] | ||
| − | PLSA is a probabilistic generative model used for topic modeling in text data. It is an extension of Latent Semantic Analysis that introduces a probabilistic framework to the topic modeling problem. | + | PLSA is a probabilistic generative model used for topic modeling in text data. It is an extension of [[Latent]] Semantic Analysis that introduces a probabilistic framework to the topic modeling problem. |
| − | * </b>How PLSA Works</b>: In PLSA, it is assumed that documents are generated through a probabilistic process. Specifically, it assumes that there are latent topics, and each document is a mixture of these topics. Each word in a document is generated from one of these topics with a certain probability. | + | * </b>How PLSA Works</b>: In PLSA, it is assumed that documents are generated through a probabilistic process. Specifically, it assumes that there are [[latent]] topics, and each document is a mixture of these topics. Each word in a document is generated from one of these topics with a certain probability. |
* </b>Applications</b>: PLSA is mainly used for discovering topics in large document collections. By analyzing the word-topic and topic-document distributions learned by PLSA, you can identify the prevalent themes or topics within the corpus. | * </b>Applications</b>: PLSA is mainly used for discovering topics in large document collections. By analyzing the word-topic and topic-document distributions learned by PLSA, you can identify the prevalent themes or topics within the corpus. | ||
| Line 23: | Line 23: | ||
LSA is a technique used for dimensionality reduction and discovering the underlying structure in a collection of documents. It's primarily used for tasks like document clustering, information retrieval, and document summarization. | LSA is a technique used for dimensionality reduction and discovering the underlying structure in a collection of documents. It's primarily used for tasks like document clustering, information retrieval, and document summarization. | ||
| − | * </b>How LSA Works</b>: LSA operates by performing Singular Value Decomposition (SVD) on a term-document matrix. This matrix represents the frequency of terms (words) in documents. SVD reduces the dimensionality of this matrix and extracts latent semantic patterns. The resulting lower-dimensional representations can help identify relationships between words and documents. | + | * </b>How LSA Works</b>: LSA operates by performing Singular Value Decomposition (SVD) on a term-document matrix. This matrix represents the frequency of terms (words) in documents. SVD reduces the dimensionality of this matrix and extracts [[latent]] semantic patterns. The resulting lower-dimensional representations can help identify relationships between words and documents. |
* </b>Applications</b>: LSA can be used for clustering similar documents, finding related documents in information retrieval, and generating document summaries by identifying key terms and phrases. | * </b>Applications</b>: LSA can be used for clustering similar documents, finding related documents in information retrieval, and generating document summaries by identifying key terms and phrases. | ||
Revision as of 09:41, 16 September 2023
Youtube search... ...Google search
- Large Language Model (LLM) ... Natural Language Processing (NLP) ...Generation ... Classification ... Understanding ... Translation ... Tools & Services
- Topic Model/Mapping
- Beautiful Soup a Python library designed for quick turnaround projects like screen-scraping
- Term Frequency–Inverse Document Frequency (TF-IDF)
PLSA is a probabilistic generative model used for topic modeling in text data. It is an extension of Latent Semantic Analysis that introduces a probabilistic framework to the topic modeling problem.
- How PLSA Works: In PLSA, it is assumed that documents are generated through a probabilistic process. Specifically, it assumes that there are latent topics, and each document is a mixture of these topics. Each word in a document is generated from one of these topics with a certain probability.
- Applications: PLSA is mainly used for discovering topics in large document collections. By analyzing the word-topic and topic-document distributions learned by PLSA, you can identify the prevalent themes or topics within the corpus.
Latent Semantic Analysis (LSA)
LSA is a technique used for dimensionality reduction and discovering the underlying structure in a collection of documents. It's primarily used for tasks like document clustering, information retrieval, and document summarization.
- How LSA Works: LSA operates by performing Singular Value Decomposition (SVD) on a term-document matrix. This matrix represents the frequency of terms (words) in documents. SVD reduces the dimensionality of this matrix and extracts latent semantic patterns. The resulting lower-dimensional representations can help identify relationships between words and documents.
- Applications: LSA can be used for clustering similar documents, finding related documents in information retrieval, and generating document summaries by identifying key terms and phrases.
Key Differences: LSA & PLSA
- LSA is primarily focused on dimensionality reduction and finding semantic patterns in documents, whereas PLSA is a generative probabilistic model designed specifically for topic modeling.
- LSA does not involve a probabilistic generative process, while PLSA explicitly models the probability of word generation from topics.
- In PLSA, the number of topics is typically a parameter to be determined, whereas LSA does not inherently model topics.
