Do you mean Quasi-experimental? I have never heard of semi-experimental. In an experiment, participants are randomly assigned to a treatment. If possible, this is the best to use because other variables can be controlled. Quasi-experimental is used when it would be unethical or impractical to use an experiment. For example, seeing the effect drug use has on high school performance. Matching should be used to account for extraneous variables when using a quasi-experimental design.
I do not know if you wish to refer to quasi-experimental designs, but if that is the case, read below.
PERRY (2008, p. 89) explains that "both experimental and quasi-experimental research designs involve manipulating the independent variable(s) and observing the change in the dependent variable(s)".
The difference between experimental and quasi-experimental designs, according to DÖRNYEI (2007), is that the latter do not use random assignment of participants to groups because of practical constraints.
DÖRNYEI (2007) - Research Methods in Applied Linguistics
PERRY (2008) - Research in Applied Linguistics: Becoming a Discerning Consumer
See also
MACKEY & GASS (2005) - Second Language Research: Methodology and Design
With reference to experimental research, participants in both treatment and control groups are randomly assigned.
The term quasi experiment is used in research and not semi-experiment. In quasi experiment, an independent variable is manipulated, but the participants are not randomly assigned.