LSI keywords in Title and description tags is as important as well as content keywords or for this matter say keyword density. But for a blog to succeed it is essential to have LSI keywords in Title and Description Tags which makes it easier for Google as well as readers to know what the content is going to talk about. When you talk about being at the top of the Page ranking you have to develop quality contents and that too with LSI keywords if present in Title and Description Tags will complete your successful efforts to be at the top of search engine. As we read about the LSI in the previous post, in this post we will understand additional uses as well as challenges to LSI.
Additional Uses
It is well understood fact that ability to work with text on semantic basis is important in latest information retrieval systems. As a result, the use of LSI has significantly expanded in recent years as earlier challenges in scalability and performance have been overcome.
LSI is being used in a variety of information retrieval and text processing applications, although its primary application has been used for concept searching & automated document categorization. Some ways in which LSI is used:
- Automated document classification
- Information Discovery
- Relationship Discovery
- Matching technical papers
- Customer Care
- Determining authorship
- Spam Filtration
- Information visualization
- Essay scoring
- Literature discovery
Challenges
As we see the additional usage of LSI we should look at the challenges of it like scalability and performance. It requires relatively high speed computational performing units as well as memory when compared to other techniques. But with introduction of new high performance processors as well as availability of cheap memory solutions, these limitation are now in control . These days applications which involves >30 million documents are easily processed through matrix and SVD computations and is not at all difficult for LSI applications.
Another challenge to LSI is difficulty in understanding optimal number of dimensions to be used for performing the SVD. In general fewer dimensions allow wider comparison of concepts in text, while higher count of dimensions enables more relevant comparison of concept. Right number of dimensions which are to be used is controlled by number of documents in collection. The past research demonstrated that almost 300 dimensions usually provide best results with a moderate size document collections and 400 dimensions is required for bulkier document collections. Also more recent independent studies indicate 50-1000 dimensions are suits depending upon size and nature of the document collection.
When you are checking the amount of variation in data post computation, SVD should be used to understand optimal dimensions to retain. The variation in the data should be viewed by mapping singular values (S) in a scree plot. Some LSI professionals select dimensionality associated with arc of the curve as cut-off point for determining number of dimensions which should be retained. Other practitioners argue that particular quantity of variance should be retained, and hence amount of gap in the data dictates proper dimensionality which is to be retained. 70% is usually mentioned as right amount of variance in the data that can be used to select optimal dimensionality for recomputing the SVD.
With this we catch hold of both the challenges as we as the additional uses of LSI and thus the importance of LSI is better understood now. Will you now use the LSI in Title and description tags? If you liked the article do share it with others on Twitter. Also if you have any ideas which can be featured in our upcoming posts you can drop in contact us form and I will be happy to look after quoting you in the post.
Tele-Columnist!