Context. Hot cores represent critical astrophysical environments for high-mass star formation, distinguished by their rich spectra of organic molecular emission lines. Nevertheless, comprehensive statistical analyses of extensive hot core samples remain relatively scarce in current astronomical research. Aims. We aim to utilize high-angular-resolution molecular line data from the Atacama Large Millimeter and Submillimeter Array (ALMA) to identify hot cores, with a particular focus on weak-emission candidates, and to provide one of the largest samples of hot core candidates to date. Methods. We propose to use spectral stacking and imaging techniques of complex organic molecules (COMs) in the ALMA-ATOMS survey, including line identification and weights, segmentation of line datacubes, resampling, stacking and normalization, moment 0 maps, and data analysis, to search for hot core candidates. The molecules involved include CH3OH, CH3OCHO, C2H5CN, C2H5OH, CH3OCH3, CH3COCH3, and CH3CHO. We classify cores with dense emission of CH3OH and at least one molecule from the other six molecules as hot core candidates. Results. In addition to the existing sample of 60 strong hot cores from the ALMA-ATOMS survey, we have detected 40 new weak candidates through stacking. All hot core candidates display compact emission from at least one of the other six COM species. For the strong sample, the stacking method provides molecular column density estimates that are consistent with previous fitting results. For the newly identified weak candidates, all species except CH3CHO show compact emission in the stacked image, which cannot be fully resolved spatially. These weak candidates exhibit column densities of COMs that are approximately one order of magnitude lower than the ones of the strong sample. The entire hot core sample, including the weak candidates, reveals tight correlations between the compact emission of CH3OH and other COM species, suggesting they may share a similar chemical environment for COMs, with CH3OH potentially acting as a precursor for other COMs. Among the 100 hot cores in total, 43 exhibit extended CH3CHO emission spatially correlated with SiO and (HCO+)-C-13, suggesting that CH3CHO may form in widely distributed shock regions. Conclusions. The molecular line stacking technique is used to identify hot core candidates in this work, leading to the identification of 40 new hot core candidates. Compared to spectral line fitting methods, it is faster and more convenient, and enables weaker hot cores to be detected with greater sensitivity.