The weights in a text spatial weights file are used as-is. No attempt is made to standardize or restandardize to accommodate elimination of bad records (e.g., missing data, null values, invalid geometry) or features excluded from the active selection set. If the input weights are row standardized, and there are bad records or an active selection set, results will be incorrect. If the weights in the text spatial weights file are not row standardized, results are correct (even with bad records or an active selection set).
If the weights in your text spatial weights file are not row standardized, there is no problem. If the weights in your text spatial weights file are row standardized, and you see a warning indicating bad records were encountered, or you want to run the analysis on a selection set, first convert the text weights to a .swm file using the Generate Spatial Weights Matrix tool. If your text file contains self weights (a weight for the relationship of a feature to itself) and you want to use those weights to run the Hot Spot Analysis tool, unfortunately, you will need to create a Self-Potential field in your input feature class with those self-weights for them to be utilized in the analysis. The conversion from table to .swm format using the Generate Spatial Weights Matrix tool removes all self weights.